|
| Greg's diary |
| January 2023 |
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Sunday, 1 January 2023 | Dereel | Images for 1 January 2023 |
| Top of page | ||
| next day | ||
| last day |
|
60 years!
|
Topic: history, general, opinion | Link here |
It's been 60 years since I started keeping a regular diary! How time flies!
60 years? Only a month ago I noted that I had completed 30 years. Why the difference?
As I mentioned at the time, I stopped keeping the diary in 1 October 1970 and didn't start again until 1 September 2000—29 years, 11 months.
But wait, there's (much) more. In their Christmas message, Eric Allman and Kirk McKusick noted the sad passing of Glenn Groenewold's husband, Wes Muchmore. They gave Wes' diary to the GLBT historical society—all 4 carloads of it! Clearly he wrote on paper, and it seems that he started long before I did, but I'm still greatly impressed. If I had kept my diary on paper, even without interruptions it would probably pretty much only fill a single bookshelf. I'm humbled.
|
Welcoming the New Year
|
Topic: general | Link here |
New Year's Day today, and a Sunday at that. Time to sleep in and recover from last night's festivities.
Oh, wait, that was in a different millennium. We haven't made it to midnight for 12 years. On the other hand, today was forecast to be hot (37°). In fact, it hit 41°. A good idea to take the dogs for a walk earlier, while it was relatively mild (only 30°).
And we weren't the only ones. Graeme was mowing his lawn, Fiona unloading hay:
|
||||||||||||||||||||||||
Somehow the old clichés no longer apply.
|
Sending the New Year's message
|
Topic: technology, general, opinion | Link here |
New Year, new New Year's message. I dread sending it: it gives me all sorts of problems with email.
Today I decided that it was no use fighting the big companies that have come into existence long after I started the LEMIS mail server nearly 31 years ago. Instead I'd relay it through Aussie Broadband's server. But:
Jan 1 09:01:08 eureka postfix/smtp[75154]: F3A442635E3: to=<foobar@wehi.edu.au>, relay=mail.aussiebroadband.com.au[121.200.0.25]:25, conn_use=4, delay=0.62, delays=0.15/0.37/0.04/0.05, dsn=4.7.1, status=deferred (host mail.aussiebroadband.com.au[121.200.0.25] said: 450 4.7.1 Error: too much mail from 121.200.11.253 (in reply to MAIL FROM command))
Jan 1 09:01:08 eureka postfix/smtp[74950]: 016DB264AAC: to=<beastie@FreeBSD.org>, relay=mail.aussiebroadband.com.au[121.200.0.25]:25, conn_use=4, delay=0.62, delays=0.16/0.37/0.03/0.06, dsn=4.7.1, status=deferred (host mail.aussiebroadband.com.au[121.200.0.25] said: 450 4.7.1 Error: too much mail from 121.200.11.253 (in reply to MAIL FROM command))
That wasn't a Christmas message, it was spam! It seems that Aussie stops accepting mail after 40 messages in a short period of time. I waited for a while for the mail queue to drain, but it stayed:
...
F12272635E5 522 Sun Jan 1 09:01:07 grog@lemis.com
(delivery temporarily suspended: connect to mail.aussiebroadband.com.au[121.200.0.25]:25: Connection refused)
bewaerbaby@gmail.com
F08C1264AE6 521 Sun Jan 1 09:01:08 grog@lemis.com
(delivery temporarily suspended: connect to mail.aussiebroadband.com.au[121.200.0.25]:25: Connection refused)
estland@crisny.org
F2740264AE9 527 Sun Jan 1 09:01:08 grog@lemis.com
(delivery temporarily suspended: connect to mail.aussiebroadband.com.au[121.200.0.25]:25: Connection refused)
lig@gmail.com
-- 101 Kbytes in 199 Requests.
OK, we can fix that. Change the mail relay back to lax.lemis.com, and off it went.
But that wasn't all. First I got this message:
grolsch@gmail.com> (expanded from <grolsch@elch.com>): host
gmail-smtp-in.l.google.com[74.125.20.26] said: 550-5.7.26 This message does
not pass authentication checks (SPF and DKIM both 550-5.7.26 do not pass).
SPF check for [lemis.com] does not pass with ip: 550-5.7.26
[54.202.1.33].To best protect our users from spam, the message has
550-5.7.26 been blocked. Please visit 550-5.7.26
https://support.google.com/mail/answer/81126#authentication for more 550
5.7.26 information. a10-20020a170902900a00b00189b279b8d3si3009433plp.45 -
gsmtp (in reply to end of DATA command)
Of course! I need to add Aussie to my SPF list.
And then there were the messages I was expecting. Bigpond (a site currently with an invalid security certificate) says:
<go.cuy@bigpond.com>: host extmail.bigpond.com[203.36.137.234] said: 558
5.7.1 6392E81D059CEBB3 Message content rejected due to suspected spam.
IB703 (in reply to end of DATA command)
Suspected spam. People, if you don't know, in dubio pro reo. And what part of my message upsets their broken spam detector? All I wrote was:
Merry Christmas from Greg, Yvonne, and myriad animals. Our New Year's
letter is at http://www.lemis.com/grog/Albums/Christmas/2022.php
All the best for the New Year
Greg and Yvonne
And then there's Outlook.com, which is not even that clear:
<mry@area.com>: host area-com.mail.protection.outlook.com[104.47.56.138]
said: 550 5.4.1 Recipient address rejected: Access denied. AS(201806281)
[CO1NAM11FT013.eop-nam11.prod.protection.outlook.com] (in reply to RCPT TO
command)
What does that mean? It could almost mean that the user no longer exists. But I'm guessing that it's either a broken spam filter or a decision that small, unknown servers are untrustworthy, even if they have been around since before Microsoft knew what email was.
So what do I do? Add the Aussie SPF records to mine, and try again to send the messages to these broken servers. If that doesn't work, I'll have to do it manually via Gmail.
What a pain these modern times are! Once email Just Worked.
|
Dragonflies!
|
Topic: animals, Stones Road house, general | Link here |
Round dusk we discovered hundreds of dragonflies flying around outside. It was too dark to get a photo, even too dark to see exactly what they were. Where did they come from? Maybe we'll find some left over tomorrow.
| Monday, 2 January 2023 | Dereel | Images for 2 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Hay!
|
Topic: general | Link here |
It's been a strange spring and summer, and Yvonne has been worrying continually about getting enough hay for the coming autumn and winter. Finally she succeeded and persuaded Fiona Drayton to sell her the rest of her hay (from directly across the road). Yvonne was looking for 200 bales, but they ended up with 220:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
That should keep her happy for a while.
|
More deep freeze problems
|
Topic: general, opinion | Link here |
Somehow I forgot to enter some foodstuffs into the database, and this morning we spent some time looking for some sliced salmon. We found it, but not before Yvonne found this:
|
|||||||||||||||||||||||
Empty plastic freezer bags. I reuse them, and they stay in the freezer to avoid having to wash them. But these had fallen down behind the baskets and the wall, where they got heated to melting point and stuck together by the defrost mechanism:
|
|||||||||||||||||||||||
That's not the first time: six months ago I paid an eye-watering $280 to have somebody come and remove a bag that had clogged up the defrosting mechanism. Admittedly it had suffered more:
|
|||||||||||||||||||||||
So: don't do that, then. But how do I store them compactly and with easy access?
|
Registering Perfectly Clear
|
Topic: photography, technology, opinion | Link here |
I've been using “Perfectly Clear“ for touching up photos fo some time now, but Yvonne doesn't. And she needs it. OK, install.
Oh. It's there. Load some photos for processing:
|
|
|||||||||
Well, I have a license. It's just not associated with this (Microsoft) user on distress. How do I enter the license key? Buy? No, that just takes me to the user manual, after starting a non-default web browser (“Internet Explorer”). Looked through the menu at the top. Nothing obvious. “Get Support”? No, that gives me the user manual too.
Trials & ActivationSimply click Activate and copy and paste or carefully type in the license number you have been provided. When you click Activate again, we will validate the license and will unlock Perfectly Clear for continual use.
But where? The illustration looks nothing like what I have, and I can't find an “Activate” button anywhere.
Much more searching, and I finally found it: it's under “About”. I suppose that's par for
the course for modern software. But I wasn't done. Yes, it accepted the license key, but
it still only allowed 5 images. First I had to guess that I had to restart
the program app for it to work.
On the positive side, of course, it's good of the maker to allow people to use the software without a license, with only the restriction of 5 images per run. And maybe this pain is due to the fact that they have discontinued “Perfectly Clear” (2 days ago), and they're producing a new product, “Radiant Photo”, at a price that I can resist. It seems that right now they don't have any product for sale.
| Tuesday, 3 January 2023 | Dereel | Images for 3 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Domain name renewal, the easy way
|
Topic: technology, general, opinion | Link here |
The domain name registration for lemis.com needs renewing. Straightforward, right?
Well, yes, in principle: go to GANDI, select lemis.com and renew.
But: I want to pay by debit card in Euros: PayPal charges too much for international
transactions. And I should check that I still have enough credit. OK, off to TransferWise to check.
What a pain this site is! Every time I go there I need to find my way around it again. But yes, I had enough € credit, just not enough USD. Transferred some money from my bank account, which of course arrived in AUD. How do I transfer it? After a lot of searching, I gave up. It's not that important: if I pay in USD and there's not enough credit, they'll choose another account (which? how do I know?) and transfer automatically under the same conditions. The only issue would be a different exchange rate, which could even be to my advantage.
OK, back to the domain name. Renewal only in AUD! Or at least, I didn't find a way to change it. OK, renew one year. Login, please. Did that. No, no, not the old, worn-out handle approach, try again with the new web interface.
OK, did that, And it changed its language to French! They wanted the card number and things, of course, also what kind: Visa, Mastercard and more. What do I know? It's a “virtual card” (in other words, just numbers). Visa maybe? Put that in and completed the transaction, slowly. And Wise had another “security” question for me: which of this list of recent transactions had I made? I didn't recognize any of them, so I clicked on “none of these”. Correct! Is that a good security question? Or does it reflect my relatively seldom use of the “card”?
Only later did I discover that it was MasterCard. What's the purpose of the question? The information is implicit in the card number.
Done! Renewed for another year. And I got a confirmatory email:
Vous recevez cet email car votre adresse a été renseignée pour recevoir les notifications qui concernent l'organisation groggyhimself chez Gandi.
So now I'm an organization all by myself!
|
More pale lilies
|
Topic: gardening, opinion | Link here |
Last year I noticed a number of new “wild” lilies in many places outside the house:
My guess was that they're self-seeded. But they were everywhere this time last year, and sure enough, here they come again, many more:
|
||||||||||||||||||||||||
|
||||||||||||||||||||||||
I can't make up my mind whether I like them or not, but Yvonne wants to use them for flower arrangements.
|
KL Hokkien Mee approved
|
Topic: food and drink, opinion | Link here |
Mail from Philip Paeps today. His partner Cat (further names to be established) comes from KL, and she found my recipe for KL Hokkien Mee. She approves, which is reassuring.
Of course, she has suggestions for improvement. The powdered ikan bilis must be there for the flavour, and so should much more garlic. I'll think about both suggestions.
|
Piccola goes walkabout again
|
Topic: animals | Link here |
It's amazing how much more healthy Piccola is looking after only two weeks of more natural diet. I had never thought that canned cat food could have had such a detrimental effect. Now she goes outside from time to time, and this afternoon she went out and didn't come back.
Problems? Well, the outside door to the laundry was open, and we have lots of new hay. Where there's hay, there are mice. Do they taste better than the food we're giving her? Hopefully she'll be back tomorrow morning.
| Wednesday, 4 January 2023 | Dereel | Images for 4 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Piccola returns
|
Topic: animals | Link here |
As expected, Piccola showed up again early this morning, apparently not as successful in her hunting as last time.
But she still doesn't overly like the food we're giving her. She left some of the (for human consumption) minced beef behind. And when Yvonne came home from shopping, I gave her some animal minced beef, fatter, and some minced chicken. The minced beef went away immediately, and she left the other two behind. OK, we can eat the remainder of the human mince that she hasn't touched yet. I wonder if this is balanced enough.
|
What's this?
|
Topic: food and drink, opinion | Link here |
While looking through the pantry this morning, I found this:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The only English text is “Recipe Mix”, “Sustainably grown” and the manufacturer. The rest is all in Tagalog. What is it? It turns out that Sinigang is a fish stew, but I can't recall such a recipe, nor buying it.
What is it? How do you use it? This time Chromium didn't help, and I had to mess around with my phone, which produced:
|
|||||||||||||||||||||||
Clearly they don't want me to read the recipe.
A bit later Yvonne confirmed that she had bought these sachets as fish broth from the local Filipino grocery (Ballarat Asian Grocery). I suppose I should try the stuff in some Malaysian dishes, though it seems that it contains a considerable quantity of tamarind, so I need to choose carefully.
| Thursday, 5 January 2023 | Dereel | Images for 5 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Installing hydra
|
Topic: technology, opinion | Link here |
I've been dawdling with installing a replacement for eureka for nearly 6 years now. Since May last year it has been the HP Z800 that I inherited from Bruce Evans. Why haven't I done anything yet?
Why, haven't I done anything yet? Yes, I've done a lot of thinking. But the delay has two main reasons:
But I'm getting there. Over the past couple of days I have been addressing the most complicated issue, /etc and startup. One thing's sure: /etc/rc.conf isn't enough. How do you update it? My solution is to create a directory /etc/rc.conf.d/ and put individual files in it for each feature. Then in the install Makefile I have something like (in fact exactly) this:
RCCONFDIR = /etc/rc.conf.d
ntp: ${RCCONFDIR}/ntp
${RCCONFDIR}/ntp: ${RCCONFDIR}
cp -p rc.conf.d/ntp ${RCCONFDIR}
cp -p etc/ntp.conf /etc
But that's only part of the problem. There's also the question of ports, some of which require entries in rc.conf. From the installation of apache24:
To run apache www server from startup, add apache24_enable="yes"
in your /etc/rc.conf. Extra options can be found in startup script.
Surely something in /usr/local/etc/ would be more appropriate? Or should I just put the info in /etc/rc.conf.d/? It's this kind of consideration that's slowing me down.
And finally (logically first): what about users? They need an entry in /etc/passwd and possibly in /etc/group. Those files can also get updated on new releases or with new ports. root is installed on all new systems, but the password is in the old /etc/master.passwd. I have a very flaky approach to that, and I'll probably leave it in place this time round. But gradually I'm getting to a point where I can attempt an installation.
And how will I structure the network? This time I think I'll leave eureka as the central hub in the network, and hydra will just be my personal workstation. This may change, of course.
|
Governing the USA
|
Topic: politics, opinion | Link here |
The USA is showing no signs of resolving its internal conflicts. They've been trying to elect a speaker for the House of Representatives for days now, and nobody has received an absolute majority of votes.
The speaker presumptive is Kevin McCarthy, but he has only received round 200 votes. He needs an absolute majority of 218 votes to be elected.
So: why not just a simple majority? The candidate with the most votes is Hakeem Jeffries, with 212 votes. Why not elect him? Sure, he's a “Democrat”, but if the Republicans can't even agree on a speaker, are they really a single party?
In passing, “218” raises a flag in my memory: error 218 meant “interrupt timeout” with Tandem.
| Friday, 6 January 2023 | Dereel | Images for 6 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
KL Hokkien Mee yet again
|
Topic: food and drink, opinion | Link here |
KL Hokkien Mee again for breakfast today. I had intended to incorporate some of the suggestions that Cat, Philp Paeps' partner, had made, but in the end I just put in Even More garlic, 15 g in this case. I had intended to put in a lot of ground ikan bilis, but I forgot. About the biggest difference was the way I fried the squid: in the house, where it produces a lot of smoke:
|
|||||||||||||||||||||||
To minimize the impact I put a fan in the hallway to the laundry:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
It worked up to a point; the rooms weren't as smoky as previously. And as the first photo shows, I was able to get the pan to a temperature that I couldn't have got any other way.
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Yes, that's Celsius (712° F).
I've been using frozen broth from chicken wings sous vide as part of the sauce, and up until today I first thawed it out and then mixed it with the other sauces. But there's no need for that. Just put it in the pan and melt it, then add the other sauce ingredients:
|
|||||||||||||||||||||||
And here the final dish:
|
|||||||||||||||||||||||
|
Ten years, still the same old procrastination
|
Topic: history, technology | Link here |
Seen in my diary for January 2013:
Once again I've been dragging my heels on my system upgrade method. In principle amd64-stable now contains all the ports I asked for and a relatively recent version of FreeBSD 9-STABLE, but I still need to customize it, and then I'll be in a position for the first upgrade. Spent some time customizing /etc/group and /etc/master.passwd; the latter contains a lot of history, user IDs and passwords from people who must at at least one time in the last 20 years have accessed the machine. Should I remove them? It's somewhat nostalgic to have the IDs there, and they won't really do much harm.
Things are still progressing slowly, but they do seem to be progressing.
| Saturday, 7 January 2023 | Dereel | Images for 7 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Building a fence
|
Topic: Stones Road house, general | Link here |
Jesse Walsh along today to continue the surroundings of the riding arena. We need a short fence, only 4 posts. Years ago CJ Ellis brought along machinery to build the fences—about 1 km of them—around the house and between the paddocks, but for 4 posts it's hardly worth the trouble. Still, it's a lot of work by hand. Towards the end of the day we were this far:
|
|||||||||||||||||||||||
| Sunday, 8 January 2023 | Dereel | Images for 8 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Sad Abutilons
|
Topic: gardening, opinion | Link here |
It's been nearly 2 years since I cloned a couple of Abutilon cuttings. One I planted in the garden, and it grew well. The other I left in the pot, and it did less well. Here they are in March last year:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
But since then things have changed. The one I planted started looking sick in about October last year. I thought it was because of the excessive rainfall, but it hasn't improved:
|
|||||||||||||||||||||||
The best that I can say is that there seems to be some new growth:
|
|||||||||||||||||||||||
The one in the pot, however, has improved, and until recently it was flowering relatively well:
|
|||||||||||||||||||||||
And then I forgot to water it:
|
|||||||||||||||||||||||
It took some time to recover:
|
|||||||||||||||||||||||
So: what have I learnt? Abutilons don't like being dry. And whatever the problem is with the Abutilon that I planted, it's not water. Once again I'm left wondering what's wrong with the soil here. We never had this kind of issue in Kleins Road.
|
COVID-19 again
|
Topic: health | Link here |
Yvonne received a Facebook message from Margaret Swan this morning: do we have any medication for sore throats? Yes, indeed, Strepsils, which have been lying around for years. It seems that she found a dead RAT in her car, and so she's alone at Chris Bahlo's while Chris is somewhere (Gumeracha?) jousting. Margaret seems to be doing well enough, but that was clearly not the weekend she was looking for.
|
Goodbye Johnbot
|
Topic: technology, opinion | Link here |
Network throughput problems again today. Uploading files slowed from 200 kB/s (a limit set by my TCP parameters, which I must fix) to about 15 kB/s. Why?
I didn't find out before it got better again, but I was able to establish that the problems weren't local. But the traceroute output was interesting for other reasons:
=== root@dereel (/dev/pts/3) ~ 4 -> traceroute www
traceroute to www.lemis.com (45.32.70.18), 64 hops max, 40 byte packets
1 eureka (192.109.197.137) 0.372 ms 0.367 ms 0.358 ms
2 radiation-tower.aussiebb.net (121.200.8.1) 27.376 ms 33.867 ms 20.654 ms
3 HundredGigE0-0-0-8.core2.vdc03.mel.ByeJohnBotThankYouForEverything.aussiebb.net (159.196.252.154) 178.964 ms 169.305 ms 183.004 ms
4 10.241.5.56 (10.241.5.56) 189.950 ms
10.241.4.156 (10.241.4.156) 190.197 ms 173.706 ms
5 be23.cfl1.nextdc-s1.syd.ByeJohnBotThankYouForEverything.aussiebb.net (180.150.1.144) 182.201 ms 196.911 ms 192.211 ms
6 10.241.13.104 (10.241.13.104) 170.963 ms 189.428 ms 186.806 ms
7 be30.bdr1.coresite-la1.lax.ByeJohnBotThankYouForEverything.aussiebb.net (202.142.143.199) 172.025 ms 179.606 ms 200.801 ms
...
12 www (45.32.70.18) 180.799 ms 174.466 ms 170.007 ms
ByeJohnBotThankYouForEverything.aussiebb.net? A bit of searching brought up this article. It seems that it refers to John Alexander, who left Aussie Broadband at the end of the month. Clearly there was no animosity involved.
| Monday, 9 January 2023 | Dereel | Images for 9 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
hydra: taking the plunge
|
Topic: technology, opinion | Link here |
I've been thinking for a long time about how to proceed with hydra, the new HP Z800 that I inherited from Bruce Evans. I have had a basic install of FreeBSD 13.1 on it for weeks, but I wasn't sure how to continue.
Today I took the plunge. And yes, it wasn't anything like what I had planned. TL;DR: even working out a strategy is an incremental process.
First there's a certain amount of bootstrapping. The install scripts are on eureka, so I have to be able to reach it. For some reason I hadn't set up the network interfaces, so I had to do that, and of course I had to mount the file systems.
That in itself was a problem, because the structure of the configuration script hierarchy shows the developments of the last 25 years. For reasons that no longer seem to make sense, I have subtrees for each system. So: currently I have the (new) install scripts on /src/Sysconfig/Install, itself a symlink to eureka:/home/src/Sysconfig/Install. To mount:
mkdir /eureka
mount eureka:/ /eureka
mount eureka:/home /eureka/home
ln -s /eureka/home/src /
At this point I have access to the source tree on eureka, so I can start building a new system:
rmdir /usr/src
ln -s /src/FreeBSD/git/stable-13 /usr/src
cd /usr/src
make -j 16 buildworld kernel
And that carried on in the background for a surprisingly short period of time, about 75 minutes. That's reassuring, anyway.
This was the one that worried me most, but in fact things didn't go too badly. I chickened out with passwd and group, and discovered that I hadn't planned to upgrade the root entry, but things seem to have worked anyway so far.
And /etc/rc.conf.d? That's such an obvious name for the directory that I discovered that it already exists, though it's empty. It's also undocumented. I wonder how it is intended to be used. Still, my method seems to work, to the point that it's not clear what use rc.conf is any more. There's nothing in it any more, but I left a couple of lines:
hostname=hydra.lemis.com
hostip=192.109.197.129
# Read files in /etc/rc.conf.d
for i in /etc/rc.conf.d/*; do
. $i
done
Installing ports consists of three parts:
Identify the ports. I created a file ports.hydra with one entry per line.
Deciding how to install them. In most cases it's as simple as
pkg install $port
Configuring the ports. There lies the rub. For those ports requiring configuration, I have separate targets. But I'm not overly happy with the situation.
That's about as far as I got today. But:
This box takes forever (well, over a minute) before it even tries to boot the OS on the disk. It seems to want to try everything else first, and I haven't found a way to tell it where to boot from.
During the boot process, ntpd wants to take to me:
[ntpd@hydra /]$
It appears to be something shell-like. But why? I can stop it, and the boot continues. But clearly I need to find a way to suppress it.
This box uses so much power, about 220 W continuously. I'm running powerd, but so far it doesn't seem to make any difference. That's 5.3 kW a day, or $1.32 per day, $480 odd a year. That's comparable to the value of the machine. I should start looking for an alternative that uses (much) less power.
root contains a number of symlinks to my home directory for things like .bshrc and .emacs. But .bashrc didn't work: by default root has /bin/csh as its shell. I thought of changing that, but no, it's probably better to leave it there. Instead, at the end of ~root/.cshrc I added:
bash
That way if bash isn't installed, I'll still have a shell.
Some of these old configuration files are ancient. In ~root/.bash_profile I found:
uuprogress ()
# display the progress of a uucp transfer from system $1
{
while [ -f /usr/spool/uucp/$1/T* ] ; do
echo `date '+%H:%M:%S';ls -l /usr/spool/uucp/$1/T*`
...
That must be over 30 years old, going back to when I set up my first mail server.
While searching for ideas, discovered a new program, etcupdate. It seems to address some of the issues I'm trying to solve here, but probably not all of them. The documentation is also quite long, so I'll put this one on the tuit queue.
Once again I ran into this irritating problem with pkg that it prompted for permission to install Every Single Port. There's a switch somewhere, but where? /etc/src.conf? No, I've been there before and left my opinions:
# Don't drop build date, dammit!
# What a silly name for the knob.
# It's about as "intuitive" as the name of this file.
WITHOUT_REPRODUCIBLE_BUILD=dammit
What does the man page for pkg-install say?
FILES
See pkg.conf(5).
OK, what does the man page for pkg.conf say? Nothing! The section is missing, and it just refers back to the indivudual programs, in this case pkg-install.
Finally I found it (it's hidden in the text): /usr/local/etc/pkg.conf. And of course the name of the knob is intuitive:
ASSUME_ALWAYS_YES = true;
During ports installation, I ran into routing errors. It looked as if the system had lost its network configuration. OK, add explicit configuration to /etc/rc.conf:
ifconfig bge0 $hostip 255.255.255.0
defaultrouter=192.109.197.137
There's plenty of time later to find out what went wrong.
Things got worse, not better. After every port it lost connection to the Global Internet. Yes, I could reinstate it, but what causes it? Mañana.
|
No carrots for Elena
|
Topic: animals, food and drink, opinion | Link here |
This evening Yvonne found some carrots that have seen better days. OK, give them to the dogs. Larissa smelt hers among the pellets, sought it out and ate it first. Elena, on the other hand, isn't so keen. She ate all the pellets and left the carrot over:
|
|||||||||||||||||||||||
Next time Lara gets them all.
|
Steak and kidney again
|
Topic: food and drink, opinion | Link here |
Steak and kidney pie (for some definition of “pie”) for dinner today. I've tried them in the “hair dryer” “air fryer” before, but they didn't brown evenly:
|
|||||||||||||||||||||||
But now we have the “bath tub”. Tried them there, with various results. I get the impression that the heat settings on the bathtub give cooler results than on the hair dryer. In the end I turned the temperature up from the initial 200° to 250°, after which they were quite nicely browned:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
But the underside wasn't so good. The dough stuck to the aluminium foil:
|
|||||||||||||||||||||||
Yvonne had some dough left over, so she made some apple pies:
|
|||||||||||||||||||||||
I cooked them at 250° from the start, and after 8 minutes they were more than done:
|
|||||||||||||||||||||||
So clearly I need more experimentation.
| Tuesday, 10 January 2023 | Dereel | Images for 10 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More hydra fun
|
Topic: technology, opinion | Link here |
Spent a considerable amount of time today consolidating yesterday's experiences. And clearly there was a lot to consolidate.
What else is there to do? Install more ports. But I couldn't find Google chrome! All I found was:
=== root@hydra (/dev/pts/0) ~ 13 -> pkg search chromium
chromium-bsu-0.9.16.1_2 Arcade-style, top-scrolling space shooter
ungoogled-chromium-108.0.5359.124 Google web browser based on WebKit sans integration with Google
What's that nonsense? Yes, I was able to install ungoogled-chromium, but that's not what I'm looking for. Some discussion on IRC, where I discovered still more undocumented details: there's a file /etc/pkg/FreeBSD.conf, which is only mentioned in passing in pkg.conf(5). And it contains information on which repository to use. By default it contains:
url: "pkg+http://pkg.FreeBSD.org/${ABI}/quarterly",
But for whatever reason (dare I even ask?) this repo doesn't contain chromium. Instead I needed to change it to
url: "pkg+http://pkg.FreeBSD.org/${ABI}/latest",
And after that I get:
chromium-108.0.5359.124 Google web browser based on WebKit
Apart from that, why does ntpd prompt during boot? I can't see any good reason, unless it's related to a lack of network interface (another issue that needs to be fixed). And rwhod doesn't seem to work. Spent some time investigating, but came to no clear conclusion. ruptime showed a single system, one of the Microsoft boxen, and not even hydra itself.
OK, how about configuring X? startx worked out of the box, but that was without the Nvidia drivers. Install that:
Jan 10 16:17:52 hydra pkg[6085]: nvidia-driver-390-390.154 installed
Jan 10 16:18:01 hydra kernel: NVRM: The NVIDIA Quadro FX 5800 GPU installed in this system is
Jan 10 16:18:01 hydra kernel: NVRM: supported through the NVIDIA 340.xx Legacy drivers. Please
Jan 10 16:18:01 hydra kernel: NVRM: visit http://www.nvidia.com/object/unix.html for more
Jan 10 16:18:01 hydra kernel: NVRM: information. The 390.154 NVIDIA driver will ignore
Jan 10 16:18:01 hydra kernel: NVRM: this GPU. Continuing probe...
Well, at least that's clear. Install the correct driver:
Jan 10 16:19:01 hydra pkg[6132]: nvidia-driver-340-340.108_3 installed
Jan 10 16:19:05 hydra kernel: nvidia0: <Quadro FX 5800> on vgapci0
Jan 10 16:19:05 hydra kernel: vgapci0: child nvidia0 requested pci_enable_io
But that was the end of the good news:
=== root@hydra (/dev/pts/1) /var 54 -> startx
xauth: file /root/.serverauth.6200 does not exist
X.Org X Server 1.21.1.4
X Protocol Version 11, Revision 0
Current Operating System: FreeBSD hydra.lemis.com 13.1-STABLE FreeBSD 13.1-STABLE #0: Mon Jan 9 12:57:21 AEDT 2023 root@hydra:/usr/obj/eureka/home/src/FreeBSD/git/stable-13/amd64.amd64/sys/GENERIC amd64
(==) Log file: "/var/log/Xorg.0.log", Time: Tue Jan 10 16:22:09 2023
(==) Using config file: "/etc/X11/xorg.conf"
(==) Using system config directory "/usr/local/share/X11/xorg.conf.d"
================ WARNING WARNING WARNING WARNING ================
This server has a video driver ABI version of 25.2 that this
driver does not officially support. Please check
http://www.nvidia.com/ for driver updates or downgrade to an X
server with a supported driver ABI.
=================================================================
(EE) NVIDIA: Use the -ignoreABI option to override this check.
(EE)
Fatal server error:
(EE) no screens found(EE)
What's that? About the best thing I can decide is that WARNING is a new term for “fatal error”. OK, I could try the -ignoreABI option, but where does that go? Off to the well-hidden documentation with the unlikely URL http://us.download.nvidia.com/XFree86/Linux-x86/185.18.14/README/appendix-b.html. No mention of an option IgnoreABI. But then, there are no leading hyphens in these options. More googling, and found various forum pages like this one, which recommended this section in the configuration file:
Section "ServerFlags"
Option "IgnoreABI" "1"
EndSection
It seems that X keeps changing ABIs, requiring changes to the X server. That in itself is surprising in a product that is a third of a century old and doesn't seem to see much evolution. But by this time there was no time to try it out.
The URL of the documentation (and also the copyright notice there) suggests that the documentation relates to a very old version of the driver. But I couldn't find anything by removing successive parts of the URL. Only much later and with much searching did I find newer versions of the Nvidia documentation. I can't find links to it anywhere, but https://download.nvidia.com/XFree86/Linux-x86_64/450.80.02/README/ could be the newest.. It still doesn't mention ignoreABI.
Apart from that, spent a bit of time investigating power consumption more carefully. The box uses a consistent 220 W odd, and powerd doesn't seem to make any change at all. Another thing that I'm going to have to chase up on. More and more I'm coming to the conclusion that this isn't the box for me. At the very least I should consider new display cards. The ones in eureka are even older, so that's not an option.
|
How I hate git, once again
|
Topic: technology, opinion | Link here |
Why couldn't I find chromium in the list of packages? There's a directory /usr/ports/www/chromium in the FreeBSD Ports Collection. What does /usr/ports/UPDATING say? Only that I haven't updated the tree in over 6 months. OK,
=== root@dereel (/dev/pts/5) /src/FreeBSD/git/ports 42 -> Log git pull
hint: You have divergent branches and need to specify how to reconcile them.
hint: You can do so by running one of the following commands sometime before
hint: your next pull:
hint:
hint: git config pull.rebase false # merge
hint: git config pull.rebase true # rebase
hint: git config pull.ff only # fast-forward only
hint:
hint: You can replace "git config" with "git config --global" to set a default
hint: preference for all repositories. You can also pass --rebase, --no-rebase,
hint: or --ff-only on the command line to override the configured default per
hint: invocation.
fatal: Need to specify how to reconcile divergent branches.
OK, git, how did that happen? I didn't do anything with it. Still, I have a choice. What does it mean? “Fast forward only” looks the most appropriate:
=== root@dereel (/dev/pts/5) /src/FreeBSD/git/ports 43 -> git config pull.ff only=== root@dereel (/dev/pts/5) /src/FreeBSD/git/ports 44 -> Log git pull
===== Tue 10 Jan 2023 11:47:03 AEDT on dereel: git pull
fatal: Not possible to fast-forward, aborting.
Grrr. That helps. Try git config pull.rebase true. Still fails: it still tries to fast forward, because that's what's in the configuration. How do I get rid of it? How do I know? Where's the configuration file? The man page is far too polite to mention that sort of thing, but from previous pain I recall a maze of little twisty config files, all different. Is there on in this directory? Yes! And it ends in:
[pull]
ff = only
rebase = false
Remove the first line and things work. And there's still no entry in UPDATING!
Why is this all so flaky?
|
Frijoles de la olla: lard?
|
Topic: food and drink | Link here |
My recipe for frijoles de la olla, the first step in frijoles refritos, includes 80 g of lard for 1 kg of beans. That creates about 3 kg of beans. To make the frijoles refritos, I take a 180 g portion of the frijoles de la olla and fry them with 15 g of lard.
The portion of frijoles de la olla corresponds to 4.8 g of lard. Why do I bother? I can put it all in when frying them. My guess is that the original recipes include the possibility that the beans will be eaten directly, in which case the lard makes sense.
Today was the day for another batch. No lard. I'll see how it works out.
| Wednesday, 11 January 2023 | Dereel | Images for 11 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Hydra: done?
|
Topic: technology, opinion | Link here |
Further work on installing hydra today. There wasn't really too much more to do. The most obvious thing was to get X up and running. In preparation connected a second monitor to the machine, and noted that that one (lower connector on the back of the machine) is the primary monitor: that's where the console appeared. So it seems that it can handle having only the second monitor connected.
As I discovered yesterday, I needed to add this to the configuration file to get X to run:
Section "ServerFlags"
Option "IgnoreABI" "1"
EndSection
Did that and tried again:
=== grog@hydra (/dev/pts/1) ~ 2 -> startx
X.Org X Server 1.21.1.4
X Protocol Version 11, Revision 0
Current Operating System: FreeBSD hydra.lemis.com 13.1-STABLE FreeBSD 13.1-STABLE #0: Mon Jan 9 12:57:21 AEDT 2023 root@hydra:/usr/obj/eureka/home/src/FreeBSD/git/stable-13/amd64.amd64/sys/GENERIC amd64
================ WARNING WARNING WARNING WARNING ================
This server has a video driver ABI version of 25.2 that this
driver does not officially support. Please check
http://www.nvidia.com/ for driver updates or downgrade to an X
server with a supported driver ABI.
=================================================================
(WW) NVIDIA: The driver will continue to load, but may behave strangely.
(WW) NVIDIA: This driver was compiled against the X.Org server SDK from commit e6ef2b12404dfec7f23592a3524d2a63d9d25802 and may not be compatible with the final version of this SDK.
(EE)
Fatal server error:
(EE) no screens found(EE)
Not exactly the success I had been hoping for. Anything in the log file?
[ 8499.072] (EE) NVIDIA: Failed to initialize the NVIDIA kernel module. Please see the
[ 8499.072] (EE) NVIDIA: system's kernel log for additional error messages and
[ 8499.072] (EE) NVIDIA: consult the NVIDIA README for details.
[ 8499.072] (EE) No devices detected.
Oh. I had rebooted hydra since the last time, and the manually loaded driver had gone away. kldload nvidia did the trick, and then the server seemed to run normally, once I had extended my fvwm configuration to recognize hydra. After that it came up with single display on both monitors:
screen #0:
dimensions: 3840x1200 pixels (1049x321 millimeters)
resolution: 93x95 dots per inch
depths (7): 24, 1, 4, 8, 15, 16, 32
At least for the time, that might not be a bad idea, though there's clearly a problem: the left-hand monitor has a resolution of 1920x1200, but the right-hand one has only 1920x1080, so the screen has 230,400 undisplayable pixels. That will go away when I connect to the correct monitors.
So at this point the system is almost usable. There are still a large number of loose ends:
In my install scripts I have a number of things to do:
Apart from that, I'm coming more and more to the conclusion that this box is only a temporary measure. Apart from the power consumption, it's also noisy.
So what do I do now? hydra only has 2 display outputs, so I can't run my entire desktop (4 screens) on it. And I can't use the cards in eureka: they're even older, and they need a different display driver. But There Can Only Be One. So in the short term I should connect two screens to each system and use x2x to connect to each of them. Yet another stopgap measure. In the mid-term I need to find a new machine.
|
Mobile phones: the smart
|
Topic: technology, opinion | Link here |
Tried to call Yvonne on the phone in town today. No answer, followed by a call back: she hadn't been able to answer the call.
So I picked up my phone, which was in landscape orientation and displaying one of its call indications: Dismiss/Accept? Turned it to portrait orientation. It dismissed without me coming anywhere near the screen.
Why does this keep happening? Is it my specific brand (Xiaomi)? Why is answering an incoming call such a minefield?
| Thursday, 12 January 2023 | Dereel | Images for 12 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Noodle experiments
|
Topic: food and drink, opinion | Link here |
For some time I've had various things in the fridge and freezer that really need to go, in particular hard dòufu and pickled chili paste:
|
|||||||||||||||||||||||
OK, how about trying to use it in a breakfast noodle dish? The chili precludes using one of my spice pastes, so let's try it from scratch.
How much chili? The pickle is surprisingly salty, so that limits it. I started with 20 g, but it didn't look like enough, so I increased it to 30 g.
| quantity | ingredient | step | ||
| 30 g | spring onions | 1 | ||
| 10 g | garlic | 1 | ||
| 10 g | ginger | 1 | ||
| 30 g | pickled chili | 1 | ||
| 75 g | hard dòufu | 2 | ||
| 50 g | cooked prawns | 2 | ||
| 60 g | pok choi | 2 | ||
| 60 g | tauge | 2 | ||
| 160 g | 5 mm rice noodles | 3 | ||
| 40 ml | water | 3 | ||
| 15 g | soya sauce | 3 |
Cut spring onions into 1 cm segments and fry with garlic, ginger and chili:
|
|||||||||||||||||||||||
Cut dòufu and gradually add other ingredients:
|
|||||||||||||||||||||||
Serve with cucumber sticks:
|
|||||||||||||||||||||||
The results? Not salty enough! I had to add the 15 g of soya sauce to adjust it. But there was far too much chili. Next time I'll try with 10 g of chili and 20 g of soya sauce and see how that works out.
Apart from that, I've decided that I don't really like the dòufu. It has a strange texture. Maybe I should use it differently (mapo dòufu maybe?).
|
rwhod works!
|
Topic: technology, opinion | Link here |
Why doesn't the version of rwhod on hydra work? I still haven't investigated. But today I discovered that distress has once again taken it into its own hands to install software updates and reboot! I noticed it because
=== grog@eureka (/dev/pts/34) ~/Photos/20230113 429 -> ruptime
distress up 15:46, 0 users, load 0.00, 0.00, 0.00
hydra down 19:27
That reminded me that I have other issues with rwhod: so far, every time distress reboots, it loses rwhod. But not today.
Why is rwhod such an issue? Programmatically it's pretty uninteresting. I think that the issue with Microsoft is probably in the installation, not the program itself, but what's wrong with the most recent rwhod on hydra?
|
More thoughts on “air fryers”
|
Topic: food and drink, opinion | Link here |
Read an interesting article in the Neue Zürcher Zeitung today, about “air fryers”. It agrees with my viewpoints, so it must be right. An interesting detail was that the term “air fryer” dates from about 2010, when Philips demonstrated the first one in an exhibition in Berlin.
But I have had one since 2009! Ah, but it wasn't called an “air fryer”. It was a convection oven, as I commented a couple of months ago. Is there really a difference?
Apart from that, the NZZ article uses the New Swiss German term « Hype » to describe the devices. It seems that social media and such are the driver for the things. And both this test and this one couldn't find any device that fulfilled all the requirements. These photos, from the article and the referenced Chip article, set the tone:
I'd be ashamed of either result.
|
Google Translate fun
|
Topic: technology, language, opinion | Link here |
I have a certain love-hate relationship with Google Translate, but I have to admit that it's getting better. But then, looking at what it did 7 years ago, that shouldn't be difficult.
So: for the benefit of the non-German speakers who might be interested in the Neue Zürcher Zeitung article above, I tried to translate it into English. I almost failed. They seem to have changed the interface to be more “modern”, and it no longer works the way I know. Neither does the method they recommend on most browsers. Only Chrome on Microsoft manages it, after much mouse pushing. Only I can't copy the URL, though what I get looks like it should be in English: https://bellevue.nzz.ch/kochen-geniessen/der-airfryer-ist-das-kuechengeraet-der-stunde-wieso-dieser-hype-ld.1719838?_x_tr_sl=de&_x_tr_tl=en&_x_tr_hl=en&_x_tr_pto=wapp.
|
More X fun
|
Topic: technology, opinion | Link here |
Didn't do any work directly on hydra today. Instead I turned my attention to another X server on eureka. Server 0 has 4 screens. From the configuration file:
Section "ServerLayout"
Identifier "Layout0"
Screen 0 "Screen0" 0 0
Screen 1 "Screen1" RightOf "Screen0"
Screen 2 "Screen2" RightOf "Screen1"
Screen 3 "Screen3" RightOf "Screen2"
InputDevice "Keyboard0" "CoreKeyboard"
InputDevice "Mouse0" "CorePointer"
Option "Xinerama" "0"
EndSection
The rest of the file describes the individual components. OK, we don't need Screen 0 or 1, so I changed it to:
Section "ServerLayout"
Identifier "Layout0"
# Screen 0 "Screen0" 0 0
# Screen 1 "Screen1" RightOf "Screen0"
Screen 2 "Screen2" 0 0
Screen 3 "Screen3" RightOf "Screen2"
InputDevice "Keyboard0" "CoreKeyboard"
InputDevice "Mouse0" "CorePointer"
Option "Xinerama" "0"
EndSection
Starting X gave me Screen 2, but not Screen 3. What went wrong? There were 5,082 lines of log file to look through, and nothing that showed any obvious issue with Screen 3. I need to compare more carefully, not helped by the fact that diff trips over the timestamp and GPU ID in things like (first server 0 log, then server 3 log):
[ 178.142] (II) NVIDIA(GPU-0): --- Building ModePool for BenQ G2420HD (DFP-0) ---
[ 178.142] (II) NVIDIA(GPU-0):
[ 178.142] (II) NVIDIA(GPU-0): Native mode timings for BenQ G2420HD (DFP-0):
[ 178.142] (II) NVIDIA(GPU-0): 1920 x 1080 @ 60 Hz
[ 178.142] (II) NVIDIA(GPU-0): Pixel Clock : 148.500 MHz
[ 178.142] (II) NVIDIA(GPU-0): HRes, HSyncStart : 1920, 2008
--------------------------------------------------
[2533927.167] (II) NVIDIA(GPU-1): --- Building ModePool for BenQ G2420HD (DFP-0) ---
[2533927.167] (II) NVIDIA(GPU-1):
[2533927.167] (II) NVIDIA(GPU-1): Native mode timings for BenQ G2420HD (DFP-0):
[2533927.167] (II) NVIDIA(GPU-1): 1920 x 1080 @ 60 Hz
[2533927.167] (II) NVIDIA(GPU-1): Pixel Clock : 148.500 MHz
[2533927.167] (II) NVIDIA(GPU-1): HRes, HSyncStart : 1920, 2008
But those lines show that the server found the configuration, and Server 3 issued no further warnings. But Server 0 had a large number of:
[ 178.145] (WW) NVIDIA(GPU-0): The EDID for BenQ G2420HD (DFP-0) contradicts itself: mode
[ 178.145] (WW) NVIDIA(GPU-0): "720x480" is specified in the EDID; however, the EDID's
[ 178.145] (WW) NVIDIA(GPU-0): valid HorizSync range (24.000-83.000 kHz) would exclude
[ 178.145] (WW) NVIDIA(GPU-0): this mode's HorizSync (15.7 kHz); ignoring HorizSync check
[ 178.145] (WW) NVIDIA(GPU-0): for mode "720x480".
So what's the issue?
| Friday, 13 January 2023 | Dereel | Images for 13 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Goodbye hydra?
|
Topic: technology, opinion | Link here |
Today I again spent more time thinking about how to complete the hydra install (peripherally including setting up the X server 3: on eureka) than actually doing anything on the machine. Somehow the noise and power consumption mean that I won't be able to use it very much. What's the alternative?
Long discussion with Daniel O'Connor and Juha Kupiainen on IRC. How about that, there are relatively fast machines available for reasonable prices:
<Doco> Ryzen 5 5500 with a lowend motherboard and 64GB of RAM is $600 (no case, PSU, HDs etc)
That corresponds to the cost of power for hydra for 15 months. How does it compare in performance? Off to CPU benchmark to compare “fast” processors. The results were surprising. The current fastest processor is the AMD Ryzen Threadripper PRO 5995WX, rated at 94571 CPUMarks (whatever that means). At $6499 it's out of my range, but it helps relativize things. It also has a massive 64 cores, 128 threads! Now that would be worthy of the name hydra.
By comparison, we have:
So: it's clear that hydra isn't worth running on a permanent basis. Maybe I can use it as a Microsoft box that only gets used now and then. A new box it is, along with a lot of questions:
Regarding the processors, built up a table of the processors of interest:
| Processor | CPU Mark | Single thread | Cores | Threads | TDP | $ | CPUMark/$ | CPUMark/W | ||||||||
| Ryzen Threadripper PRO 5995WX | 94571 | 3290 | 64 | 128 | 280 | 6499 | 14.55 | 338 | ||||||||
| Ryzen 9 7950X | 63513 | 4314 | 16 | 32 | 170 | 589.99 | 107.63 | 374 | ||||||||
| Ryzen 9 7900X | 52076 | 4319 | 12 | 24 | 170 | 448.99 | 115.97 | 306 | ||||||||
| Ryzen Threadripper PRO 5955WX | 49550 | 3345 | 16 | 32 | 280 | 1300 | 38.12 | |||||||||
| Ryzen 9 5950X | 45881 | 3467 | 16 | 32 | 105 | 499 | 91.95 | |||||||||
| Ryzen 7 7700X | 36458 | 4274 | 8 | 16 | 105 | 349 | 104.48 | |||||||||
| Ryzen 9 3900X | 32741 | 2940 | 12 | 24 | 105 | 375 | 87.31 | |||||||||
| Ryzen 5 7600X | 28749 | 4212 | 6 | 12 | 105 | 274 | 104.93 | |||||||||
| Ryzen 5 5600X | 21937 | 3356 | 6 | 12 | 65 | 179.40 | 122.28 | |||||||||
| Ryzen 5 5600 | 21536 | 3248 | 6 | 12 | 65 | 179.94 | 119.68 | |||||||||
| Ryzen 5 5600G | 19847 | 3180 | 6 | 12 | 65 | 109.71 | 180.90 | 305 | ||||||||
| Ryzen 5 5500 | 19487 | 3049 | 6 | 12 | 65 | 119.40 | 163.21 | 300 | ||||||||
| Core i7-4771 | 7087 | 2180 | 4 | 8 | 84 | 129.74 | 54.63 | 84 | ||||||||
| Xeon X5660 | 6080 | 1380 | 6 | 12 | 95 | 54.07 | 112.44 | 64 |
One thing of interest would be the CPUMark to TDP ratio. But after starting to calculate it, I found that it was pretty meaningless. It does show, however, how badly the Xeons perform in this area. On the other hand, the comparison suggests that the Ryzen 5 5600 might be a better bet, or potentially the 7600X. The real question is how much power the chips really use. At least, though, I now have a better overview.
In passing, these CPUMarks are a moving target. 7½ years ago the i7 4771 in eureka was rated at 9933 CPUMarks. Now it's only 7087. PassMark acknowledge that the values can change, but the bar code that they publish doesn't quite agree with my records:
|
|
|||||||||
Have they changed their criteria in that time?
|
Have gun, will clean
|
Topic: general | Link here |
Victoria Caulton along today to clean the house. She certainly looks the part:
|
|||||||||||||||||||||||
And no, the portable induction cooker in her left hand isn't part of her equipment. I lent it to her to try out before buying a “real” induction cooker.
| Saturday, 14 January 2023 | Dereel | Images for 14 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Summer coming?
|
Topic: gardening | Link here |
Gradually it's getting warmer—today the temperatures hit 38°—but the grass is still green, and it's growing like fury. And conveniently Jesse Walsh wasn't able to come today, so it looks like we're going to have to mow it ourselves. And there are still some tiny flowers visible in the nature strip in front of the house:
|
|||||||||||||||||||||||
|
|
|||||||||
The Hibiscus rosa-sinensis “Uncle Max” has recovered, sort of, but it clearly took a bad hit:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
But the first flowers are coming:
|
|||||||||||||||||||||||
|
How old is that chili?
|
Topic: food and drink, photography | Link here |
Years ago, when I still made kimchi regularly, I bought a large quantity of gochugaru (고춧가루):
|
|||||||||||||||||||||||
And much of it is still there. I use a small quantity for sprinkling on the eggs of huevos rancheros, but it will last for many years more. Clearly it's time to put it in a jar.
But when did it expire? The price is still legible, but the date isn't:
|
|||||||||||||||||||||||
Once again, time for a detailed photo:
|
|||||||||||||||||||||||
That's with the best accentuation I could get. The problem is that increasing contrast also increases the structure of the paper. But at least it's legible, though I feel I should have been able to do better.
|
Indian food again
|
Topic: food and drink, language, opinion | Link here |
Somehow we don't eat much Indian food any more. Part is Yvonne's surprising disinterest for it. About the only thing we eat is chicken tanduri. And that's what we ate tonight.
It wasn't straightforward. First, we had no “roti” (really “wraps” for non-Indian food). So do we have any atta? After a bit of searching found this:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Clearly authentic, even with a price in Rupees:
|
|||||||||||||||||||||||
But what does the description mean? “Aashirvaad shudh chakki atta 0% Maida 100% Atta”? Google Translate helpfully identifies it as Gujarati and “translates” it into English as “Aswarwad Shudh Chakki Ato 0% Mainda 100% Ato”. Clearly Maida is undesirable, and I know what atta means, but the rest?
Anyway: the reason I stopped making my own chapatis was because it's a lot of work. But now I have a tortilla press. How hard can it be? Pretend it's a tortilla and press between paper:
|
|||||||||||||||||||||||
But that's superficial. The details look different:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Much flour needed to avoid it sticking. But then, surprisingly, it worked well:
|
|||||||||||||||||||||||
The chicken was another matter. A year ago I had established that I should first cook them in a microwave oven at 600 W for 10 minutes, then grill them at 230° in the “hair dryer” air fryer for 5 minutes. But it was clear this time round (and apparently not last time) that the heat in the microwave oven was far too much, and this time the meat wasn't browned sufficiently:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
So next time I'll greatly reduce the microwave exposure.
| Sunday, 15 January 2023 | Dereel | Images for 15 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Piccola returns
|
Topic: animals | Link here |
Piccola is improving greatly since the change of diet, and she has put on a lot of weight. She goes outside again. Previously I had thought that she just didn't like being chased by the dogs, but now I'm coming to the conclusion that she wasn't feeling very well. What a difference the diet makes!
She spent the night outside again last night. On return, strangely, she wasn't hungry. She must have found something of interest out there, and that can only benefit our hay stores.
The current diet of minced beef has a couple of disadvantages: first, it needs to be kept frozen until we feed her, so we need to thaw the correct quantity in advance. And we feed her on top of the dryer in the laundry:
|
|||||||||||||||||||||||
And the beef is stickier than the cat food she's used to, meaning that there's a lot of mess left behind when she's finished:
|
|||||||||||||||||||||||
|
Minor garden work
|
Topic: gardening | Link here |
Reading back through my diaries of years gone by, it's amazing how much worse things look than, say, ten years ago. Things really looked nice then. By comparison, there's almost nothing of interest here.
One thing that is clearly suffering is the last remaining Fuchsia in a hanging basket by the house entrance. It's been there forever, so maybe it has just reached end of life. But maybe it just needs repotting. Did that, and it currently looks like this:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
System configuration, next step
|
Topic: technology, opinion | Link here |
More thoughts on a replacement for hydra today. On IRC, discussed the amazing difference between component pricing and complete systems. Yesterday I had found a system similar to the one that Daniel O'Connor described on Friday. Daniel came out under $1000; the offer I got was for $2,259. Why?
Well, there's the case and the power supply, and some of the details don't quite match. But somehow there's still a big gap. Before I compare oranges with oranges (no Apples needed here), I need to decide what kind of display card(s) I need, and as always, that's painful. One option seems to be a refurbished card; currently there's a NVIDIA Quadro P600 available with 4 outputs for $129. Why refurbished? What can fail on a card like that? In any case, I need to investigate video cards yet again. I'm not looking forward to it.
Another interesting discussion: what power supply? Juha Kupiainen is still using a cheap power supply for his always-on machine. I pointed him at my article on the subject of 10 years ago: cheap power supplies use more power.
How much? Based on my calculations at the time, the power supply has saved me about $1,150. But running eureka in the same time has probably cost me about $3,700 in power!
| Monday, 16 January 2023 | Dereel | Images for 16 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Bathing the dogs
|
Topic: animals | Link here |
Yvonne decided that today was the day to bathe the dogs. Well, in fact she had been planning it for some days, but I wanted to get it on video. And she kept putting it off until it was stinking hot, in this case round 30°.
The dogs weren't happy: they were afraid of the hose, and it took us until the second dog, Elena, before we found out how to calm them down:
|
Bloody flies!
|
Topic: general, animals, opinion | Link here |
Somehow the flies are particularly bad this year. It makes walking (or bathing) the dogs a misery. But Yvonne found a solution:
|
|||||||||||||||||||||||
|
|
|||||||||
|
A use for OI.Share?
|
Topic: technology, photography, opinion | Link here |
I've grumbled about OI.Share at length in the past. It seems to have been written by a team that ignores the rest of the camera development. Insists on JPEG images. Downloads images in a different format from normal. Can only connect to one camera, and if it is configured to connect to another, it forgets all about the first. Insists on being an AP, thus ensuring that you need a separate operation to move the images to a real computer. And everything is slow, probably because of hardware restrictions. In general, useless.
But for one minor detail: if you use it in remote control mode, it resets the camera clock to the phone time (presumably NTP adjusted). So I can set the clock more accurately than my old Olympus OM-D E-M1 Mark I allows.
But how long does that take? Today I tried it. With the E-M1 Mark I, it took me a total of 3 minutes, 20 seconds, including finding and starting the app, waiting what seems to be an eternity for the camera to connect, answering silly questions, waiting another eternity for the camera to connect.
The Olympus OM-D E-M1 Mark II has clock menu settings that include seconds. How long does that take? 25 seconds, along with much less fiddling. It seems that even this function of OI.Share is barely worth the trouble.
| Tuesday, 17 January 2023 | Dereel | Images for 17 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
hydra, finally
|
Topic: technology, opinion | Link here |
So it's clear that hydra will only be a short-term solution for my “desktop”. It's too weak and too power-hungry. But it's still worth setting up to refine my installation procedures so that the next system (what am I going to call it?) will be easier to install.
The current state of affairs was carried over from last Thursday: I couldn't get X on eureka to display only on the right two monitors. Fix that first.
The config file that I had was derived from the configuration for server 0, with just the screen information for screens 2 and 3. And only screen 2 worked. OK, make a new configuration file with all the configuration info for screens 0 and 1 removed. That also came up with only one screen—screen 1!
But there's this program nvidia-settings that is supposed to handle this kind of layout question. Tried that. Yes! I can delete screen 0. But not screen 1. With a little more messing around, I was able to disable screen 1. Save config file? No, it wants /etc/X11/xorg.conf, which doesn't exist. A little more messing around and I was able to save it as /etc/X11/xorg.confxxx, and later rename to /etc/X11/xorg.3.conf. And how about that, it worked!
How? I haven't looked yet. I have bigger fish to fry. First, connect hydra to the left hand 2 monitors. That was much more work than I thought: hydra is some distance from my desktop, and it's good like that, since it makes for less noise. Fortunately, one of the DVI cables was long enough, but the other wasn't. Finally located a longer cable behind lagoon, Yvonne's computer. Replace with the short cable. Didn't work. It took something like 10 minutes before I realized that the large pin had been bent to one side.
Finally! Four monitors, two on hydra (hydra:0.0 spread across both screens), two on eureka (eureka:3.0 and eureka:3.1). It works!
Well, mostly. There are a surprising number of things that need attention. First, why does ntpd prompt? Looking at the process with ps was interesting:
PID TT STAT TIME COMMAND
1449 v0 I 0:00.00 _su -m -c sh -c "/usr/sbin/ntpd -b -g -p /var/db/ntp/ntpd.pid -c /etc/ntp.conf -f /var/db/ntp/ntpd.drift -g"
What's _su? Some variant of su, clearly. But the real thing is that it's visible at all; once ntpd has finished its shell, it goes away. Spent some time looking at the sources without coming up with anything obvious. But then I discovered that I still didn't have a default router. Somehow these modifications to /etc/rc.conf didn't go as smoothly as I had intended: I had lost the defaultrouter entry. Fix that, in the process removing everything from rc.conf except the loop that read in the files from /etc/rc.conf.d/, reboot, and it worked! So did rwhod, for reasons that I will probably never understand: on Thursday it wouldn't work even when the network was correctly configured.
In the process, managed to confirm the boot time for hydra: 1 minute, 45 seconds, much of it spent waiting for net boots to time out.
Then more minor fun with X. I don't have a functional keyboard on hydra: I access the server with x2x from eureka. But once again this sequence in the .xinit didn't get executed:
kbdcontrol </dev/ttyv0 -r 250.46
xmodmap keyboard/sun-map
nvidia-settings --load-config-only
On closer examination, one reason is obvious: I only execute it on eureka. I've added it to the config for hydra, and we'll see if it works next time I start X. That's not certain: I've had problems with this for three years on teevee and tiwi, and I haven't found a way round them. Put the commands in .xinitrc and they have no effect. Execute them in an xterm and they work fine.
Apart from that, there appear to be issues with .Xdefaults: the font sizes are different, the frames for xterm and Emacs are too big, some of the background colours are wrong, and the shell bindings I set 2 months ago are gone. Mañana.
And then there's firefox.Two months ago I had discovered that I could no longer use mutt to invoke a firefox instance on the same screen but from a different system. It had to be running on the same system. That has changed now: I can't invoke firefox at all. It always tries to start a new instance (and then complains that it's already running). What has changed there? Off googling, but all hits assumed that I was trying to invoke something from a running firefox. Another mañana.
And then there's an issue that I've been dragging behind me for ever: non-ASCII fonts. For example, the Korean page for Kimchi displays like this:
|
|||||||||||||||||||||||
On Microsoft it does the right thing:
|
|
|||||||||
This one does on eureka as well, because I have gone to some trouble to install the fonts. But I still don't have Devanagari, for example. I need to find a way to install all normal fonts. Maybe—for once—Microsoft could set an example there.
While looking at these things, noted that I have somehow lost my central logging entry in /etc/syslog.conf:
*.* @eureka
Put that in for all systems.
And of course, this installation gives me the opportunity to investigate hydra's power consumption. It's now closer to 280 W, and the office is noticeably warm. I shouldn't wait too long for the replacement.
|
The council taketh away, the council giveth
|
Topic: Stones Road house, general, opinion | Link here |
A few days ago we received a letter from the Golden Plains Shire Council: our property value had been reassessed. Clearly yet another sting in the tail from the riding arena. How much? They gave the new value, but not the old one, so Yvonne had called and asked for details. Nobody there to answer the call, of course, but today a Kirsty (presumed spelling) called and explained. Increased by $30,000, which seems reasonable. No effect on our rates. What, really? No, not this year, only next year. I suppose we're going to have to live with that.
On the other hand, it seems that somebody forgot to tell us that, as pensioners, we are entitled to a rebate on our council rates (or, as Kirsty put it, we were sent an application form 6 (why not 8?) years ago, and we never returned it). She'll send me another one, so possibly we'll gain a little rebate on the ever-climbing rates.
| Wednesday, 18 January 2023 | Dereel | Images for 18 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Modern tech nightmare
|
Topic: technology, general, opinion | Link here |
After listening to the news at 7:00 this morning, turned over, went to sleep, and dreamed. I was going through a toll tunnel by car, and I had forgotten to get my payment card out in advance. Never mind, at the toll booth at the other end, pull into a bay beyond the end and get it out.
But something went wrong (I forget what), and it took a while to get back to the toll booth. Tried to talk to the person behind the thick glass, but I couldn't understand what he was saying. Tried to get him to come to the side where I could hear him, but he refused.
Back to the car to get my phone. Where's the car? There were many more there, and some were covered in snow. Gave up and off looking for a way home by foot, going through grimy, dark, dreary neighbourhoods seemingly without a sky that I had only ever seen in dystopic TV programmes. Which direction was I going in? I guessed north-east, but without a phone I couldn't tell. I just didn't want to go towards Geelong, since I was already south of Dereel.
Skip through things that I can't recall. Now I was considerably north of Dereel, maybe even in Europe, and I headed north along a village street towards our office (Tandem, I think) at the end of a row of buildings on the other (west) side of the road. But something went wrong, and I got lost off to the east. Ended up in a dark, wet but green neighbourhood, though outside it had been daytime. Finally found my way back through building sites and to almost where I had got lost. Time to call people and explain where I had been for the last 2 days. But my phone wasn't behaving—it had somehow installed a new user interface, with tiny fonts and boxes round all the texts, and I could neither read nor understand it.
Round about here I woke up: 7:45. Had I really been dreaming this modern-day horror for half an hour?
So: what did the dream mean? The tunnel (or something structurally similar) crops up from time to time in my dreams. What's different here is that I was unprepared for the toll booth. Where's my card? In the Good Old Days I would have paid cash and wouldn't need anything special for this tunnel. And the thick protective glass reminds me of the kind of painful communication I so often have at reception desks.
And the gloomy neighbourhoods? A reflection on modern dystopic TV programmes, the kind that so many people seem to like, and which I find horrible.
And the phone user interface? Too close to the updates that I have had recently, and also the illegibility of the battery charge indication on enzian, the Xiaomi Redmi 9A. The home page looks like this, compared to the Xiaomi Redmi 9T on the right:
|
|
|||||||||
|
|||||||||||||||||||||||
At first sight they don't look very different, but comparing the charge display is a different matter. Here roughly double the size of the display on the phone:
|
|
|||||||||
|
|
|||||||||
It's just about readable here, but on the phone and with my old, tired eyes, it's as good as illegible. In fact, the stuff of nightmares.
|
More hydra progress
|
Topic: technology, opinion | Link here |
Gradually I'm getting hydra to a point where it's useful. An interesting point in separating the server functions (eureka) from workstation functions (hydra) is memory usage. When I boot eureka, it fills up its memory within a day or two, and before long I'm dipping into swap. Here the current situation:
last pid: 41875; load averages: 0.92, 0.91, 0.90 up 36+02:51:59 12:10:11
554 processes: 2 running, 528 sleeping, 6 stopped, 18 zombie
CPU: 23.8% user, 0.0% nice, 2.6% system, 0.2% interrupt, 73.4% idle
Mem: 6959M Active, 21G Inact, 2881M Wired, 346M Cache, 1655M Buf, 61M Free
Swap: 20G Total, 15G Used, 5271M Free, 74% Inuse
But on hydra I have:
last pid: 8002; load averages: 0.48, 0.31, 0.34 up 1+22:07:08 12:10:33
176 processes: 1 running, 175 sleeping
CPU: 0.1% user, 0.0% nice, 0.1% system, 0.0% interrupt, 99.8% idle
Mem: 1867M Active, 13G Inact, 482M Laundry, 3455M Wired, 1222M Buf, 29G Free
Swap: 80G Total, 80G Free
Will things stay that way? Based on my experience with eureka, I had planned to put 64 GB of memory into the new machine. Will 32 GB maybe be enough?
And there's still the question of power consumption. I currently have 5 machines in my office and another 2 in the rest of the house, all powered on all the time:
=== root@hydra (/dev/pts/9) ~ 9 -> ruptime
bde up 74+18:49, 0 users, load 0.13, 0.11, 0.08
dereel up 36+02:55, 0 users, load 0.03, 0.14, 0.15
dischord down 1+03:18
distress up 7+17:22, 0 users, load 0.00, 0.00, 0.00
eureka up 36+02:55, 21 users, load 1.04, 0.91, 0.89
hydra up 1+22:08, 1 user, load 0.40, 0.30, 0.33
lagoon up 35+19:48, 0 users, load 0.18, 0.30, 0.37
tiwi up 25+03:17, 1 user, load 0.14, 0.15, 0.10
How much power does that use? bde is a laptop, dereel, dischord, distress, lagoon and tiwi are ThinkCentres, which I hope don't use much power, maybe 100 W idle. I've established that hydra uses between 220 W (idle) and 280 W (active). And 10 years ago I established that “eureka” would use round 150 W with the efficient power supply that I subsequently purchased. But that was a different eureka. To do it right, I need to measure the consumption, and that means a reboot. Still, I have an order of magnitude: 2 machines (eureka and groggy (the provisional name of the new machine, for want of something better)) using 150 W each and 5 machines using 100 W each. The laptop gets lost in the noise, but that's still 800 W or so. Time to think about hibernation. And of course when this is all over, I can turn dereel off most of the time.
Other issues that I established today:
Mail works, almost. The biggest issue I have isn't with mail itself, but with mutt. Creating a message in flyspell mode gives me part of this message (it's all in the Emacs *Messages* buffer):
Error enabling Flyspell mode:
(Can’t find Hunspell dictionary with a .aff affix file)
ispell-find-hunspell-dictionaries: Can’t find Hunspell dictionary with a .aff affix file
Clearly a missing port, but which?
/etc/crontab was missing local additions, which amount to nightly cleanup tasks. Added them.
I still have issues with ssh keys. I thought that they were related to old versions of ssh, but I get:
=== grog@hydra (/dev/pts/1) ~ 4 -> ssh freefall
agent key RSA SHA256:S7sZHLcY4dgw53/rF70vrScdPuGef3enHdJzuYA1WDo returned incorrect signature type
grog@freefall.freebsd.org: Permission denied (publickey).
Disconnected at Wed 18 Jan 2023 08:41:17 AEDT=== grog@hydra (/dev/pts/1) ~ 5 -> ssh ffm
agent key RSA SHA256:S7sZHLcY4dgw53/rF70vrScdPuGef3enHdJzuYA1WDo returned incorrect signature type
Last login: Fri Jan 6 06:26:27 2023 from 121-200-11-253.79c80b.mel.nbn.aussiebb.net
FreeBSD 11.1-STABLE (GENERIC) #0 r326456: Sat Dec 2 13:06:23 UTC 2017
Welcome to FreeBSD!
...=== grog@dereel (/dev/pts/9) ~ 1 -> ssh freefall
Last login: Fri Dec 23 06:42:49 2022 from 121-200-11-253.79c80b.mel.nbn.aussiebb.net
FreeBSD 14.0-CURRENT #0 main-n259006-a6346c02f64: Sun Nov 6 02:50:08 UTC 2022 root@build-14.freebsd.org:/usr/obj/usr/src/sys/CLUSTER
Welcome to FreeBSD!
...
What version of ssh? hydra should have the latest and greatest. And it does:
=== grog@eureka (/dev/pts/48) ~ 6 -> ssh -V
OpenSSH_6.6.1p1, OpenSSL 1.0.1p-freebsd 9 Jul 2015=== grog@hydra (/dev/pts/5) ~ 55 -> ssh -V
OpenSSH_9.1p1, OpenSSL 1.1.1s-freebsd 1 Nov 2022=== grog@dereel (/dev/pts/12) ~ 1 -> ssh -V
OpenSSH_9.0p1, OpenSSL 1.1.1q-freebsd 5 Jul 2022
Also, as the output shows, ffm (also very down-rev) reports the incorrect signature, but accepts it anyway. So where is this problem hiding?
x2x works well, but it's not perfect. If I have a window that extends to the very left of eureka:3.0, I can't get to hydra: I have to move the window a couple of pixels to the right And for reasons I need to investigate, the Compose key and the row of F keys above the main keyboard no longer work. But that will go away when I have my screens on one machine.
Still, it's a lot of progress. I can use it like this and prepare for a cleaner install of groggy (or whatever I decide to call it).
|
System configuration, 30 years ago
|
Topic: technology, history, opinion | Link here |
I've had this scrap of paper behind my monitors for some time:
|
|||||||||||||||||||||||
It dates from 14 September 1991, and it's the I/O slot configuration for two machines, adagio and (I think) grave, two of my oldest Unix boxes. This was while I was still at Tandem, and my guess is that adagio was an Intel 80386 with 8 GB running Interactive UNIX System V/386, and grave might have been a smaller machine running SCO Xenix System V. It's not clear why I crossed out “Grave” without a replacement. Possibly, like now, I had decided that I didn't like the name, but didn't
Somehow this sentence trails off here. Maybe “but didn't come up with a better name”?
The details are things that we wouldn't even think about today: boards and IRQs. But there's stuff there that I had forgotten about. Both machines had SCSI interfaces (Adaptec 1520 and 1542A), an X.25 interface card, scanner, mouse adapter. And big disk, 310 MB. I had forgotten what the Wangtec PC-02 was: it was a QIC-02 tape controller. How times change!
|
Salt and pepper squid?
|
Topic: food and drink, opinion | Link here |
For some reason, Yvonne likes the idea of “salt and pepper squid”, basically breaded squid. I have no idea where the term “salt and pepper” fits in. Today she brought these back from shopping:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The cooking instructions were unconvincing: cook in convection oven for 20 to 25 minutes at 200°. That with squid! You'd end up with rubber. Checking my diary confirmed: my previous experiment, deep fried, was still tough.
Tried it in a deep frier again, as long as I dared, about 4 minutes. Surprise, surprise: it wasn't completely cooked. But it was boring, and not very satiating. Hopefully Yvonne also has had enough.
| Thursday, 19 January 2023 | Dereel | Images for 19 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More mobile phone pain
|
Topic: technology, opinion | Link here |
Reading my diary entry for a year ago shows how things haven't changed. I had difficulty findin a way to change the time before an incoming call on an Android phone diverts (to voice mail). I needed to call the support line to find out, and he gave me the code **61*101**30. But I also had a reference to a more detailed discussion 4 years ago, and there I had different codes, even different between Android and iOS. In this case, after substituting, the code should have been **61*101*11*30. There's also the code *#61# to check.
It's only coincidence that 61 is also the international code for Australia. But what's that 11? No idea. Tried it out on Yvonne's phone (which proved to be set to answer after only 15 seconds) and got:
|
|||||||||||||||||||||||
OK, that makes sense. What about my phone, hirse?
|
|
|||||||||
Now what does that mean? And why the difference? Another thing that I'll have to follow up.
|
Daily hydra fun
|
Topic: technology, opinion | Link here |
Into the office this morning. No mail about the overnight backup job. What happened? Ran the cleanup job manually:
dump -2uf - / | nice zstd -T8 > /dump/hydra-FreeBSD/2/root.bz2
DUMP: WARNING: should use -L when dumping live read-write filesystems!
DUMP: Date of this level 2 dump: Thu Jan 19 09:10:55 2023
DUMP: Date of last level 0 dump: the epoch
DUMP: Dumping /dev/da1p2 (/) to standard output
...
That looks fine, but clearly the dump hadn't run before. Off to check the cron log:
Jan 18 21:00:00 hydra /usr/sbin/cron[5778]: (root) CMD (/home/local/bin/cleanup)
That looks fine. But where's the output?
Watched the dump progress with iostat:
tty da0 da1 cd0 cpu
tin tout KB/t tps MB/s KB/t tps MB/s KB/t tps MB/s us ni sy in id
0 1134 5.0 21746 106.6 0.0 0 0.0 0.0 0 0.0 1 3 4 1 92
0 1428 5.0 25348 124.2 4.0 2 0.0 0.0 0 0.0 1 3 3 1 93
...
Over 20,000 transfers per second! Aren't these SSDs wonderful? Well, be that as it may: this was a perfectly normal magnetic disk. Where did those transfer rates come from? Was the disk controller lying? After all, it does claim to be SCSI, when in fact it was IDE. Some discussion on IRC, during which Juha Kupiainen asked for the output from diskinfo. Here the relevant parts. First /dev/da1, the SSD:
=== grog@hydra (/dev/pts/13) ~ 2 -> diskinfo -t da1
Seek times:
Full stroke: 250 iter in 0.016239 sec = 0.065 msec
Half stroke: 250 iter in 0.016996 sec = 0.068 msec
Quarter stroke: 500 iter in 0.079407 sec = 0.159 msec
Short forward: 400 iter in 0.044783 sec = 0.112 msec
Short backward: 400 iter in 0.043068 sec = 0.108 msec
Seq outer: 2048 iter in 0.308084 sec = 0.150 msec
Seq inner: 2048 iter in 0.130764 sec = 0.064 msec
Transfer rates:
outside: 102400 kbytes in 0.747658 sec = 136961 kbytes/sec
middle: 102400 kbytes in 0.746164 sec = 137235 kbytes/sec
inside: 102400 kbytes in 0.746195 sec = 137230 kbytes/sec
The seek times are far better than magnetic disks, of course, but those transfer rates don't look like anything to write home about. What about /dev/da0, the magnetic disk?
=== grog@hydra (/dev/pts/13) ~ 1 -> diskinfo -t da0
Seek times:
Full stroke: 250 iter in 4.400060 sec = 17.600 msec
Half stroke: 250 iter in 3.411145 sec = 13.645 msec
Quarter stroke: 500 iter in 3.948668 sec = 7.897 msec
Short forward: 400 iter in 1.694292 sec = 4.236 msec
Short backward: 400 iter in 2.840787 sec = 7.102 msec
Seq outer: 2048 iter in 0.138454 sec = 0.068 msec
Seq inner: 2048 iter in 0.136790 sec = 0.067 msec
Transfer rates:
outside: 102400 kbytes in 0.552272 sec = 185416 kbytes/sec
middle: 102400 kbytes in 0.645471 sec = 158644 kbytes/sec
inside: 102400 kbytes in 1.002823 sec = 102112 kbytes/sec
Those times all look about what I had expected. But:
=== root@hydra (/dev/pts/14) ~ 13 -> time dd if=/dev/da1p3 count=100000 of=/dev/null
100000+0 records in
100000+0 records out
51200000 bytes transferred in 5.591914 secs (9156078 bytes/sec)=== root@hydra (/dev/pts/14) ~ 14 -> time dd if=/dev/da1p3 count=10000 bs=256k of=/dev/null
10000+0 records in
10000+0 records out
2621440000 bytes transferred in 19.034236 secs (137722363 bytes/sec)
The transfer rate should be independent of the block size, but here it isn't. So maybe this is really an issue with the disk controller.
During the discussion,, Daniel O'Connor came up with another tool, smartmontools, which I must investigate in more detail.
Other random issues:
Mail is still not working correctly. A couple of people I know had birthdays round today: Rusty Russell and Kirk McKusick. Sent messages to both of them, but didn't get a delivery confirmation. It seems that I had an old email address for Rusty, and the external mail server (mail.lemis.com) got a bounce message that included the new address. But for some reason I didn't get a message.
Kirk was different: the message never left hydra, though I can't see where it went. In particular, it didn't get stored anywhere except in the outbox. What's going on here? Eric's revenge?
| Friday, 20 January 2023 | Dereel | Images for 20 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Stuck bone!
|
Topic: animals | Link here |
In the mornings we give our dogs a lamb bone each. But today Larissa came back with the bone still in her mouth:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
We've seen that before with Leonid: the bone gets stuck in their back teeth, and they can't dislodge it. For some reason I can't find any photos of Leo like that, but he always let us remove it wit no issues. Lara was less happy—this time—and had to be held down while we removed it.
|
More KL Hokkien Mee experiments
|
Topic: food and drink, opinion | Link here |
Yet more experiments with KL Hokkien Mee today. Last time I used more garlic, which really didn't make much difference, but I forgot the ikan bilis. This time I didn't.
How much? Lots! I was going to put in 20 g, but in the end 12 g seemed enough:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
I also cooked the squid less long, though just as hot, with the result that there was less smoke in the house.
|
|||||||||||||||||||||||
And the result?
|
|||||||||||||||||||||||
The biggest difference was that the powdered ikan bilis stuck to the pan and needed some removal (boil up with hot water). It didn't significantly improve the flavour, so I think that, once again, I'll leave my recipe the way it was. I'm left with a nagging feeling that the real problem is that my sense of taste is fading.
|
Lawn in summer
|
Topic: gardening | Link here |
Mowing the lawn in (our part of) Australia is a spring thing. In the winter it's green but doesn't grow. In summer and autumn it's brown and doesn't grow. Only in the spring does it need mowing.
But that's in a normal year. This year we have had record rainfall in the spring, and in the last couple of days we've had another 20 mm. The result is long, spring-like grass:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
I wonder how much longer this will go on.
|
More hydra fun
|
Topic: technology, opinion | Link here |
I've switched to reactive mode on the hydra installation. I'll repeat it all from scratch when I install groggy (or whatever I decide to call the new machine), and in the meantime I've found what I think is a solution for my concerns with port installations: the basic installation (from packages until proof of the contrary), and an installation of the related configuration files. All I need is to fill in the details.
Other random issues:
My cleanup job didn't run again last night. Why? /etc/dumpdates showed no change. Off searching, and found that my mail to root had not been diverted. That's another missing configuration file, /etc/aliases. And the message from cleanup was:
dump -2uf - / | nice zstd -T8 > /dump/hydra-FreeBSD/2/root.bz2
DUMP: WARNING: should use -L when dumping live read-write filesystems!
DUMP: Date of this level 2 dump: Wed Jan 18 21:00:00 2023
DUMP: Date of last level 0 dump: the epoch
...
DUMP: dumping (Pass IV) [regular files]
zstd: error 25 : Write error : Permission denied (cannot write compressed block)
DUMP: Broken pipe
DUMP: The ENTIRE dump is aborted.
dump -2uf - /home | nice zstd -T8 > /dump/hydra-FreeBSD/2/home.bz2
mkdir: /dump: Input/output error
mv: rename /dump/hydra-FreeBSD/2/home.bz2 to /dump/hydra-FreeBSD/2/home.0.bz2: Permission denied
/home/local/bin/dodump: cannot create /dump/hydra-FreeBSD/2/home.bz2: Permission denied
...
DUMP: estimated 286627056 tape blocks.
DUMP: Broken pipe
DUMP: The ENTIRE dump is aborted.
What's that all about? I've seen bogus permissions issues with NFS before, but not this many at once.
I'm still having web browser problems. firefox still (barely) runs across the network, but it crashes all the time. And I can't find a way to get chrome to set font sizes individually for each type of font, with the result that monospace displays far too small. Once again I wonder: surely this must hit other people too? But all the web hits on chrome set font size can be summarized as RTFM.
On the positive side, the system is much more responsive. That's presumably because there's less paging involved.
|
Understanding MMI
|
Topic: technology, opinion | Link here |
Yesterday I had pain trying to set timeouts for diversion to voicemail on our mobile phones. How do I find out what these codes mean? When I check the status, I get a fleeting “MMI Code Started” message before the status display. What does MMI mean?
With a bit of searching discovered that it's short for Unstructured Supplementary Service Data. Or maybe it's a superset of USSD, but the page is too polite to explain the TLA. Still, the fields in the command **61*101*11*30 mean:
BS is the type of bearer service, some valid values are:
| Saturday, 21 January 2023 | Dereel | Images for 21 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Garden flowers in mid-summer
|
Topic: gardening, opinion | Link here |
Jesse Walsh along today to mow the lawn, coincidentally a month after the equinox again and thus time for more garden flower photos. Sometimes I wonder why. I spend lots of money for gardeners to keep the place tidy, but the result—no fault of the gardeners—is really anything but satisfactory.
We've been here for nearly 8 years now, and there's precious little to show for it. The best I can say is that nothing much has died in the last month. But here's a 7-year-old Ginkgo biloba and a 5 year old Box Elder, both still alive, both well under 1 m high:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
The Corymbia ficifolia that we bought 3½ years ago (to replace the previous one that died) is also not dead yet. In fact, it has some indication that it might be preparing to flower (at a time when the trees are normally in full flower):
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
My carefully coddled Hibiscus rosa-sinensis “Uncle Max” isn't dead yet, though I missed the two flowers that it has produced so far this season:
|
|||||||||||||||||||||||
And the Camellia japonica that we brought with us from Kleins Road is still looking really sick, as it has done since we planted it here:
|
|||||||||||||||||||||||
Nearby, the Leucadendron is also looking unhappy, but in this case it's because its pot got flooded, and there's hope that it will recover:
|
|||||||||||||||||||||||
And the Trachelospermum jasminoides that has been here for hearly 8 years isn't dead yet either:
|
|||||||||||||||||||||||
For all that and all the money I spend on gardening help, the place is a mess:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Why do I bother? It wasn't always like this. Ten years ago the garden was looking good and attractive. Now it's just frustrating.
|
The road ahead
|
Topic: health, general, animals, opinion | Link here |
Yvonne has barely recovered from her last accident, one that had the potential to cripple her for life. That's at least the fourth accident she has had in the last five years, two of which involved spinal damage and required a stay in hospital. In the course of that time we established that she has osteoporosis, and that any riding would require careful precautions, including an inflatable jacket that would cushion her in the case of any fall.
And in spite of that, the last accident happened in almost ideal conditions: she was in the arena, soft sand underneath her, she fell off, the vest inflated—and still she had spinal damage. Fortunately she recovered from it, but that was just luck.
And now she wants to ride again! She'll be 74 in March, if she lives that long. The next accident could be fatal, or it could end up with her in a wheelchair. Then she will definitely stop riding. I'm not the only person to try to persuade her not to ride any more. How can I succeed where others have failed? Probably I won't be able to. I don't understand: she sees so clearly when other people are doing dangerous things, but for herself it doesn't seem to count.
So: time to prepare for a life after Yvonne. What do I do? Looking back, when I met her, I was living in a relatively comfortable house which was big enough for both of us, but not for the horses: the plot was only about 640 m² in size. Since then, all our properties have been chosen to accomodate horses. No Yvonne, no horses. So I sell this property and move into something small again.
But where? Not in a city. I haven't lived in a city since 1979, and even then it was just a suburb. And the thought of lots of people fills me with horror. I think I'd probably go back to South Australia and find somewhere to the east of Adelaide. Without a garden, probably.
| Sunday, 22 January 2023 | Dereel | Images for 22 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Email formatting, yet again
|
Topic: technology, general, opinion | Link here |
How do you write a good email message? Clearly the point is to communicate, so a good message is one that communicates well. But many people have never learnt that. 25 years ago I wrote an article “Internet mail” with recommendations on how to communicate, and a little later I wrote a document that has now become part of the FreeBSD web site: How to get Best Results from the FreeBSD-questions Mailing List. In part it recommends:
Format your message so that it is legible, and PLEASE DO NOT SHOUT!!!!!. We appreciate that a lot of people do not speak English as their first language, and we try to make allowances for that, but it is really painful to try to read a message written full of typos or without any line breaks.
Do not underestimate the effect that a poorly formatted mail message has, not just on the FreeBSD-questions mailing list. Your mail message is all people see of you, and if it is poorly formatted, one line per paragraph, badly spelt, or full of errors, it will give people a poor impression of you.
When answering a question,
Include relevant text from the original message. Trim it to the minimum, but do not overdo it. It should still be possible for somebody who did not read the original message to understand what you are talking about.
Use some technique to identify which text came from the original message, and which text you add. I personally find that prepending “>” to the original message works best. Leaving white space after the “> ;” and leave empty lines between your text and the original text both make the result more readable.
Put your response in the correct place (after the text to which it replies). It is very difficult to read a thread of responses where each reply comes before the text to which it replies.
All pretty standard stuff. Except that it seems to have fallen on deaf ears. Much of this comes from the fact that every man and his dog now use email, and they haven't learnt to communicate. In addition, they use toy programs that don't offer adequate editing capabilities. It's the exception to see a well-formatted message nowadays.
So in general, I shut up. But today I followed a message on FreeBSD-questions where one reply included:
> PS Also bugging me is dreadful endless top-posting, tail-quoting Microsoft
> corporate-email-style mess.
Yes, I can understand that. But the response was worse:
> Sorry but blame google and/or gmail for the top posting.
> I use whatever their defaults are and that's not going to change
> unless they change it.
What? Does Gmail really impose this kind of restriction? Off to check, and yes, it seems that they do.
Spent a bit of time cooling down, then off to check. Ha, ha, only joking. On a page entitled Have better email conversations [sic], there are instructions to “Reply inline to email”. But they don't exactly rub it into your face.
In passing, the “top poster” pointed me to RFC 1855, dated October 1995, which has some suggestions about “Netiquette”. I can't see anything directly relevant to the discussion, but maybe that in itself is of interest.
O tempora! O mores!
|
Sweet and sour “shrimp”
|
Topic: food and drink, opinion | Link here |
Today was the first day of the year of the water Hare (or is that rabbit? People can't seem to agree). No rabbit or hare to be had, so I chose a recipe that I had seen recently on The Woks of Life.
That's an interesting web site. They have some interesting idea and background, and the recipes don't look bad. But the presentation! There's so much advertising that you can hardly find the content. And of course it's all US American, though they do offer limited conversion to metric if you ask nicely.
So this recipe is called “Sweet and sour shrimp”, though clearly what they're using are prawns. Here their photo:
And yes, the framing of the photo is original.
So how do I interpret the recipe? Here the version partially converted to sane units:
For the shrimp:
▢ 16 peeled and deveined shrimp (16 to 20 size; 16 shrimp at this size equals about 12-16 ounces/340-450g)
OK, that means prawns with an average weight of 25 g or so.
▢ neutral oil (such as vegetable, canola, or peanut oil, for frying)
▢ 70 g all-purpose flour
▢ 1/8 teaspoon baking powder
What's that? How much is that?
▢ 20 g cornstarch
in other words, cornflour
▢ 1/4 teaspoon salt
How much is that?
▢ 1 pinch turmeric powder (optional)
Why turmeric? It's not Chinese.
▢ 1/8 teaspoon white pepper
▢ 1/4 teaspoon sesame oil
▢ 120 ml cold seltzer or club soda
What's that? I really have no idea
For the sweet and sour sauce:
▢ 30 g red onion (cut into 1-inch pieces)
▢ 35 g red bell pepper (cut into 1-inch pieces)
capsicums, right?
▢ 35 g green bell pepper (cut into 1-inch pieces)
▢ 30 g ketchup
▢ 125 g canned pineapple chunks
▢ 180 ml pineapple juice from can
▢ 80 ml water
▢ 45 ml red wine vinegar
Chinese red wine?
▢ 40 g sugar
▢ 1/4 teaspoon salt
▢ 15 g cornstarch (mixed into a slurry with 2 tablespoons water)
Do I trust the quantities for the sauce? No. The balance between sugar and vinegar is crucial, and I already have it in my sweet and sour fish recipe.
In addition I discuvered that we didn't have any tinned pineapple, so I faked it with other ingredients from the sweet and sour fish recipe. I also only had red capsicum. Ultimately I ended up with:
| quantity | ingredient | step | ||
| 200 g | raw prawns | 1 | ||
| 43 g | flour (initially 35 g) | 2 | ||
| 10 g | cornflour | 2 | ||
| 1 g | salt | 2 | ||
| 2 g | sesame oil | 2 | ||
| 60 g | water | 2 | ||
| 25 g | red onion | 3 | ||
| 80 g | red capsicum | 3 | ||
| 40 g | bamboo root | 3 | ||
| 25 g | tomato paste | 4 | ||
| 100 g | water | 4 | ||
| 40 g | Chinese wine | 4 | ||
| 17 g | sugar | 4 | ||
| 50 g | vinegar | 5 | ||
| sesame oil | 5 |
Mix the batter ingredients, coat the prawns with it, and deep fry at 165° for 2 minutes. Here's the first half:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Not what I expected, so I added more flour to the batter and tried frying in a pan, as the recipe had asked. The interesting thing is that the temperature dropped from 165° to 115° almost immediately. And the result (left half of the last photo) looked no different:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Fry the vegetables.
Mix in the sauce ingredients and bring to the boil. Refry the prawns for another 2 minutes:
|
|||||||||||||||||||||||
Put the prawns on a serving dish. Add the vinegar to the sauce, mix and bring back to the boil. Pour over the prawns, add coriander:
|
|||||||||||||||||||||||
Serve with rice.
And the results? Not bad, but I had made the mistake of omitting the soya sauce from the sauce, and not making up the salt content. The batter was a complete disaster; next time I'll get Yvonne to try something. Also, it somehow wasn't as sweet or sour as I had intended. But the sweet/sour balance was right. Comparing the all-important ratio of vinegar to sugar, I have a ratio of 3 parts vinegar to one part sugar; this recipe has a ratio of 1.11:1.
Looking at the original, which has twice the quantities I had, I already have much more vinegar and a little less sugar. Next time I'll use 50% more, which for these quantities meaans 75 g vinegar and 25 g wine, compared to 22.5 g vinegar and 20 g sugar in the original recipe.
And of course it looked nothing like the photo in the original recipe.
| Monday, 23 January 2023 | Dereel | Images for 23 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
More hydra stuff
|
Topic: technology, opinion | Link here |
I've paused my installation of hydra—again—while I think out the next step in ports installation. But gradually it's becoming important. As usual, broken web browsers are the main driver, though I also need to find a reliable way to install ports.
Breakage in invoking firefox has forced me to use chrome instead. That still works, but it has even less of an idea of X than firefox. There can only be one! I want to be able to have a browser window on all screens, but chrome will only give me one. And it may not even be visible! The worst thing is when I'm reading mail on tiwi and I try to display a web page:
Created new window in existing browser session.
But I don't see anything. That's because the browser window is two rooms away on eureka! What short-sightedness!
On the other hand, I still don't have a solution to the firefox bug. It looks like I'm in for a rocky time.
And then there's stupidity like truncating URLs. When I go to, say, http://www.lemis.com/grog/diary-jan2023.php#20, firefox gives me:
|
|
|||||||||
It's clearly too polite to repeat the http:// and just hints at it with the crossed-out person emblem before the rest of the URL. But chrome does it even worse: it drops the www!
|
|
|||||||||
That's stupid! There's nothing to stop me from having two different web servers, www.lemis.com and lemis.com, located on different machines in different parts of the world. Maybe I should do so, just like the “Why backslashes are not slashes” page that I wrote decades ago. But to my horror I discover that both firefox and chrome change the URL why\backslashes\are\not\slashes.html to why/backslashes/are/not/slashes.html. So, through sheer stupidity, they have made themselves unable to display that page! It wasn't always so: the ancient version of firefox on eureka can access the page.
And then today I ran into the dreaded “stuck Mod key” problem that I've seen before, in fact through the ages. Previously it was Mod2, but today I ended up with a stuck Mod5 on hydra. Could this be a bug in x2x? Anyway, why not just disable it?
Oh:
=== grog@eureka (/dev/pts/55) ~ 20 -> DISPLAY=hydra:0 xmodmap -e 'clear mod5'
xmodmap: please release the following keys within 2 seconds:
Print (keysym 0xff61, keycode 111)
Super_R (keysym 0xffec, keycode 116)
xmodmap: please release the following keys within 4 seconds:
Print (keysym 0xff61, keycode 111)
Super_R (keysym 0xffec, keycode 116)
xmodmap: please release the following keys within 8 seconds:
^C
Clearly that's not going to work, but it proves that if I switch to the console, I can run it. Only it doesn't reset the status, so in the end I had to shoot down X to recover. But now I have this in my .xmodmap:
clear Mod1
clear Mod2
clear Mod3
clear Mod4
clear Mod5
The other strange thing is that when I restarted X, the displays were reversed: the left-hand monitor had the right-hand half of the display, and vice-versa. How could that happen? I hadn't changed the configuration file at all. In any case, it's as simple as changing the connectors. And strangely now the Compose key works. I suspect that that's a coincidence.
|
Maggots?
|
Topic: animals, general, opinion | Link here |
Victoria Caulton along to tidy up today, and she found this in the broom:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
I wonder where they came from. The broom was next to one of the freezers the whole time. Have we a collection of flies under it?
Later her daughter Gia came along and got to know our animals, in particular Piccola:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
| Tuesday, 24 January 2023 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Browser insights
|
Topic: technology, opinion | Link here |
Continued trying to understand why I can't connect to an existing firefox instance on hydra, though it works with chrome. Bug in firefox? It looks like it. But in such cases it's good to check with the “standard”: Microsoft.
And with Microsoft that problem doesn't exist, though you can't just enter
C:\Users\grog> firefox -new-tab www.lemis.com
That's far too easy. Instead you need to enter
C:\Users\grog> c:\Program Files\Mozilla Firefox\firefox.exe -new-tab http://www.lemis.com
But that's the Microsoft that we know and hate.
So: why doesn't it work on FreeBSD? Upgrade: firefox: 108.0.2,2 -> 109.0_1,2. Not surprisingly, no change.
Discussed it on IRC. Juha Kupiainen tried it out on FreeBSD. It worked for him. What version? 109.0_1.2.
So it's my problem. But then Daniel O'Connor came up with an idea: Dbus. What's that? One of the nameless packages that I found that I had to install over 10 years ago, along with hald, but for which I had found no documentation. It's still running on eureka, but for hydra I tried it without. And how about that, X worked just fine.
But Daniel told me that it appears that firefox communicates with itself via dbus. When he invokes it from a shell, as we had been doing, he sees messages go by.
OK, what is dbus, anyway?
=== grog@hydra (/dev/pts/5) ~ 44 -> man dbus
No manual entry for dbus
Isn't that typical? Well, almost. man -k shows a whole lot, of which dbus-daemon is the most likely. It points to this description, which I haven't read yet. But the answer seems to be that I need it running. How do I start it? RTFM is one possibility; the other is to add this entry to /etc/rc.conf.d/:
dbus_enable=YES
After that, off to work on my ports install scripts. I had tried to think of various ways of merging port installation, including relieving the guesswork about the real name of the port, and the configuration files. In the end I've decided on a triplet port description: name, typical installed file, and divergent name where it exists. For example, there is no port called ImageMagick; instead I need to choose between ImageMagick6 and ImageMagick7.
And the configuration files? Put them in a directory corresponding to the port name. Then I can generate a Makefile entry like:
hugin: /usr/local/bin/hugin
(cd hugin && make)
That leaves it up to the Makefile in hugin to do the right thing with the configuration files.
|
Why backslashes are not slashes, revisited
|
Topic: technology, opinion | Link here |
Yesterday I was horrified to find that both firefox and chrome are unable to access /why\backslashes\are\not\slashes.html: they change the backslashes into slashes. And it seems to be new: the version of firefox on eureka does it correctly. So does Apple's “Safari”.
Ah, but that's an invalid URL, says somebody on IRC. Really? Off to check RFC 2396. No, it's allowed, just deemed “unwise”. I can agree with that.
| Wednesday, 25 January 2023 | Dereel → Ballarat → Dereel | Images for 25 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Feeding Piccola
|
Topic: animals, opinion | Link here |
Piccola is looking a lot healthier since we changed her diet, but she still makes a mess of her feed. It seems that she has difficulty with the consistency of the beef mince. I've been thinking about it, and last night I mixed in a bit of water to see if that would help.
No. She didn't like that at all, and left it behind. Gave her some fresh meat, which she ate voraciously.
|
Another planned power outage
|
Topic: Stones Road house, general, technology, opinion | Link here |
Powercor announced a planned outage last week, only 5 weeks since the last one. And again they informed me at length by snail mail, mobile phone messages and email.
This one was relatively short, only 2 hours between 11:00 and 13:00. And the sun was shining, the PV battery was almost fully charged. What could go wrong?
The inverter, again! That's exactly what happened last time. It turned off for the best part of a second, taking down all computers but nothing else. Damn!
As it happens, I was just typing in yesterday's diary, and I had got as far as mentioning starting dbus, requiring a reboot. And there it was!
The good news: yes, with dbus firefox works as expected. So I can get rid of Chrome, which makes life much easier. But why didn't I know that? Nobody told me, and there was no error message when I tried to attach.
Apart from that, there was the usual pain bringing up the systems, and I was done in 30 minutes, though Powercor was even faster, bringing the power back in 18 minutes. But why does the inverter cut out when the power goes away? A sudden spike in an already too-high voltage level? At the time of failure the voltage was 250.5 V, barely in the recommended range. High time to finally ask Powercor to replace the transformer. In the meantime, I reconnected the office UPS, which I had removed because it kept beeping, presumably because of the overvoltages. And that still doesn't exonerate the inverter, which should simply have disconnected from the mains.
|
Dentist's appointment and shopping
|
Topic: health, general, opinion | Link here |
Into Ballarat for a six-monthly dentist's appointment, which took very little time but cost at least as much as usual. Mario recommends Peter Apostolopoulos, the periodontist, conveniently just across the road. Leela did too, so it looks like I'll be seeing him some time in April.
On to the Fruit Shack for random purchases, and discovered to my horror that they no longer have Jack Hua noodles, nor anything similar. It seems that they have changed their offerings, and now there seem to be many more Korean foods, unfortunately none that I need. In particular, they didn't have Bibimjang (sauce for Bibimbap).
Then to The Good Guys to look for a new kitchen mixer. I shouldn't have: even simple things need research. I could get a cheap one for round $60, or spend up to $500 for things that consisted mainly of buzzwords. And when I was almost ready to buy something, I couldn't find a corresponding package. The whole time no salesperson came near me. Good Guys, you can do better than that.
|
Powercor: the sting in the tail
|
Topic: technology, opinion | Link here |
Back home, put my camera in the office and saw on a monitor:
Boot:
eureka was down! How did that happen? Another power outage? Brought it back up and was just investigating things when I saw a panic message. Damn programs that pretend to be kernels and scare me unnecessarily.
But no, it went on:
Dumping 1300 out of 32658 MB: (CTRL-C to abort) (CTRL-C to abort) (CTRL-C to abort) ..2%..12%..21%..31%..41%..51%..61%..71%..82%..92%
And yes indeed, eureka had paniced. Not for the first time: nothing else was affected, and there had been no power issues. There was also no panic dump. Between bouts of copious swearing, investigated things. What could it be? Hardware issues? Possible, but given the timing, it sounded more like file system corruption. Which file system? /home is most likely. Apart from backup disks, eureka has:
Filesystem 1048576-blocks Used Avail Capacity Mounted on
/dev/ada0p4 39,662 33,025 3,464 91% /
/dev/ada0p2 39,662 13,447 23,042 37% /destdir
/dev/ada0p5 3,705,520 2,557,655 851,423 75% /home
/dev/ada1p1 7,629,565 6,833,736 719,533 90% /Photos
OK, fsck. /home has journalled soft updates, of course, but if there's corruption, that doesn't help much. So:
|
|||||||||||||||||||||||
How about that, /var/tmp! And the time was round about when I left for Ballarat. Things continued, of course—fsck takes about an hour on this file system—and I also got:
|
|||||||||||||||||||||||
After that, just reenable journaling:
|
|||||||||||||||||||||||
And then, how about that, this time I did get a crash dump:
|
|||||||||||||||||||||||
That's comforting, because it's now clear that the panic was related to the corruption that fsck corrected. Hopefully that'll be the end of the issue until I get the new machine up and running.
In addition, hydra had logged:
Jan 25 13:16:48 hydra kernel: bge0: link state changed to DOWN
Jan 25 13:16:48 hydra syslogd: sendto: No route to host
Jan 25 13:16:54 hydra kernel: bge0: link state changed to UP
Jan 25 13:18:08 hydra kernel: nfs server eureka:n/homfe:s noste rrveesrpo ndinge
Jan 25 13:18:08 hydra kernel: u
Jan 25 13:18:08 hydra kernel: reka:/home: not responding
Jan 25 13:18:08 hydra kernel: nfs server eureka:/home: not responding
Jan 25 13:20:36 hydra kernel: nfs server eureka:/: not responding
Jan 25 13:35:00 hydra syslogd: sendto: Host is down
No idea why the network interface went down, but it seems that we still don't have atomicity in logging errors. I thought this sort of thing went waway decades ago. Still, the timing is interesting: the corruption of inode 3308807 appeared to occur at 13:16, and eureka was unresponsive by 13:18.
In passing, it's been years since I analysed a dump, but the summaries that are produced automatically are quite useful.
| Thursday, 26 January 2023 | Dereel | Images for 26 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Bibimjang
|
Topic: food and drink, opinion | Link here |
Yesterday I was unable to find any bibimjang (sauce for bibimbap):
|
|||||||||||||||||||||||
How difficult is it to make? Off looking and came up with this recipe, conveniently requiring nothing that I didn't already have. It looked a little dubious, so I didn't make much:
| quantity | ingredient | step | ||
| 20 g | gochujang | 1 | ||
| 10 g | sesame oil | 1 | ||
| 4 g | sugar | 1 | ||
| 11 g | water | 1 | ||
| 5 g | sesame seeds | 1 | ||
| 4 g | vinegar | 1 | ||
| 4 g | garlic | 1 |
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
How was it? Closer to the pre-prepared sauce than I would have expected, but definitely better. That's one to keep.
In the process, discovered that the standard Korean way of eating bibimbap is to mix all the components together before eating. That seems a pity, and I think I'll deviate from the practice. But what a lot of work it is!
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
|
How secure is my password?
|
Topic: technology, opinion | Link here |
I've grumbled about various stupid password rules in the past, but somehow others have found out too. There's this xkcd cartoon, which I have seen in the past:
And it points to this page, which claims to be able to guess how easy it is to guess the difficulty of cracking a password. It's certainly interesting, though I have my doubts about its accuracy. It seems that the password “Nospac3s” takes an hour to crack, while “No spaces” takes a week. Thus the requirement to have a digit and no spaces, no doubt.
On the other hand, just putting in the URL of the page as a password is much better. That would take “3 tresvigintillion years”, whatever that means. The one that I really use would take 17 years to crack. That should be enough.
|
More power fail consequences
|
Topic: technology, opinion | Link here |
Into the office this morning. No mail since yesterday evening! More mail configuration pain? A quick look at /var/log/maillog showed:
Jan 25 21:13:47 eureka postfix/cleanup[31862]: warning: 1822A2635C8: write queue file: No space left on device
Where did that all go? I should have the best part of 10 GB free on the file system. Off to tidy up, and then check: the backup for eureka should have gone to an external disk mounted on /backups. Did it? No.
/backups is normally not mounted. It gets mounted specifically to perform a backup of eureka. What's on the root file system under /backups? A partial backup of eureka, of course. The mount failed because the file system wasn't clean, and for some reason the backup script didn't fail as a result.
|
Fighting the flies
|
Topic: animals, general | Link here |
It's summer, and the flies are getting on our nerves again. Last week I got a veil to keep them out, and today Yvonne followed suit:
|
|||||||||||||||||||||||
|
New noodles
|
Topic: food and drink | Link here |
While shopping at the Fruit Shack yesterday I had looked for a number of noodles—and drawn a complete blank. Hopefully that's not a permanent situation. In the end I bought these rice noodles, which are packed in 45 g (or, as the packaging claims, 90 g) baskets:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Are they really rice noodles? This transparent appearance suggests beans.
|
Latina tortellini
|
Topic: food and drink, opinion | Link here |
Yvonne bought some tortellini while shopping yesterday:
|
|||||||||||||||||||||||
625 g uncooked for 4 people. That sounds far too much. Cooked they were 980 g, and we ate 430 g today. Given Yvonne's sparrow-like appetite, the estimate of 4 portions is surprisingly correct. And they didn't taste bad, something in itself: two dishes which get my thumbs up today.
| Friday, 27 January 2023 | Dereel | |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Piccola's appetite
|
Topic: animals, food and drink, opinion | Link here |
Piccola has been eating much more lately, and she's looking so much healthier than a month ago. But I still can't decide how much she wants to eat, and I've been logging her consumption.
Another thing that I had noticed was the mess she made, leaving lots of bits of food over:
|
|||||||||||||||||||||||
But that stopped a couple of days ago. Now she eats every last piece, and lots of it. In the last 48 hours she has eaten 569 g of essentially pure beef, 284 g per day. That compares to the 200 g of pellets per day that we gave to Larissa until recently (I've since increased it to 300 g, but it's not clear that will stay that way). The change corresponds with a new delivery of beef. Can it be that she prefers it?
The effects on Piccola are clear. She now has normal weight and seems much more active. I'm still having difficulty believing that ALDI could sell such dangerous cat food.
|
More Microsoft pain
|
Topic: technology, opinion | Link here |
Some years ago I discovered that I could suspend a Microsoft “Windows” machine to RAM and save a lot of power, and I still do so with dischord.lemis.com. But that was “Windows” 7, and it's out of date. Now on distress.lemis.com I'm using “Windows” 10, itself apparently on the way out, and I can't find a way to get it to suspend automatically. There's a SHUTDOWN command that will “hibernate” (suspend to disk), but that has to be invoked from a command prompt. Still, that's worth it.
Except that it doesn't work! Lately I've discovered that the machine will shut down for a few hours, and then it restarts for no obvious reason. I suspect it is some undocumented Microsof behaviour, maybe something like a cron job.
What a pain this Microsoft is!
| Saturday, 28 January 2023 | Dereel | Images for 28 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Cat food issues: insights?
|
Topic: animals, opinion | Link here |
Piccola continues to eat voraciously, while Larissa seems to find 300 g per day too much; we'll put her back to 250 g from tomorrow.
One thing that I've noticed is that Piccola started eating more since Wednesday, when she got a fresh batch of meat. It looks very different, here on the right:
|
|||||||||||||||||||||||
So maybe the other stuff doesn't taste as good. Tough. She'll have to eat it anyway.
|
More video download problems
|
Topic: technology, multimedia, opinion | Link here |
One of the first things I do every morning is to download new videos from German TV providers. I use an old version of MediathekView because the newer versions are so Linux-centric that I have difficulty installing them. This is also an obstacle in my upgrade to hydra.
But lately I've been having difficulties with some series: I get poorly defined “permissions” error messages and have to download them manually. And then today a whole lot of older episodes of „Notruf Hafenkante“ cropped up—all only in low resolution and without subtitles. In addition there was a new episode of „Der Staatsanwalt“, also low resolution. On the positive side, I had no difficulties downloading the episodes.
But why should a brand-new episode be only in low resolution? Off to download „Der Staatsanwalt“ manually. Surprise, surprise:
=== grog@tiwi (/dev/pts/5) /spool/MediathekView 472 -> ls -l Staatsanwalt/ /spool/Series/Staatsanwalt/18/
/spool/Series/Staatsanwalt/18/:
-rw-r--r-- 2 grog home 76392 26 Jan 00:04 18-04-Wein_und_Mord.deu.vtt
-rw-r--r-- 2 grog home 2380483160 28 Jan 13:39 18-04-Wein_und_Mord.mp4
-rw-r--r-- 2 grog home 119246 28 Jan 10:23 18-04-Wein-und-Mord-20230203-201500.srt
Staatsanwalt/:
-rw-r--r-- 1 grog home 579832497 28 Jan 10:44 Wein-und-Mord-20230203-201500.mp4
-rw-r--r-- 1 grog home 119246 28 Jan 10:23 Wein-und-Mord-20230203-201500.srt
-rw-r--r-- 1 grog home 242742 28 Jan 10:23 Wein-und-Mord-20230203-201500.ttml
-rw-r--r-- 1 grog home 557 28 Jan 10:23 Wein-und-Mord-20230203-201500.txt=== grog@tiwi (/dev/pts/5) /spool/MediathekView 473 -> mpid Staatsanwalt/Wein-und-Mord-20230203-201500.mp4 /spool/Series/Staatsanwalt/18/18-04-Wein_und_Mord.mp4
Staatsanwalt/Wein-und-Mord-20230203-201500.mp4 VIDEO: [H264] 960x540 24bpp 25.000 fps 1183.5 kbps (144.5 kbyte/s)
/spool/Series/Staatsanwalt/18/18-04-Wein_und_Mord.mp4 VIDEO: [H264] 1920x1080 24bpp 50.000 fps 5270.1 kbps (643.3 kbyte/s)
Most videos are in 720p, but this time I got 540i from MediathekView and 1080p from a direct download. Why? Is this because of an old version of MediathekView? It seems unlikely. In addition, something that I had noted before: the subtitles from MediathekView have extensions .ttml and .sub, while the direct downloads are usually only .deu.vtt or .de.vtt. The former appears to be the standard .sub format, while the latter can be either that or .ttml. Clearly MediathekView has done some needed tidying up—when it works.
|
More Hugin pain
|
Topic: technology, photography, opinion | Link here |
As part of migrating to hydra, processed my weekly house photos on hydra. That started surprisingly well; the biggest issue was that I only had two monitors, while I used to do the processing on eureka server 1:, which spans all four monitors. And I no longer have that: the second monitor is no longer connected to eureka at all.
And of course it all looks different, due to some library or another. And the fast preview window comes up blank. To get it to work at all, I need to click on the “start fast preview” button. And for reasons I haven't investigated, it lists file names in reverse order. But apart from that, it seemed to work.
Oh. This image came up with a big underexposed smudge at the border between two images about 25% of the way from left:
|
|||||||||||||||||||||||
OK, try it manually. That came up with all sorts of problems. The “align images” function hung in cpclean. I left it running for a while and did something else, but after 1½ hours I have to accept that it's not going to finish:
PID USERNAME THR PRI NICE SIZE RES STATE C TIME WCPU COMMAND
17930 grog 1 108 5 1657M 111M CPU6 6 89:02 99.99% cpclean
So I used the PTO file generated from my batch script. But for some reason the arena on the left and the rear part of the driveway on the right are overexposed. We can fix that: change the parameters for the underlying images. But that hung hugin completely, and I had to shoot it down. Both on eureka and on hydra. And when I tried to stitch it on hydra, it failed. From the log:
Blending images...
enblend: internal error: unhandled command line option
What's that? There's nothing in that specific image that uses different options, and it worked on eureka. Somehow hugin has come up with multiple issues at once, and I don't know where to go from here.
| Sunday, 29 January 2023 | Dereel | Images for 29 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
The smart of phones
|
Topic: technology, general, opinion | Link here |
I've grumbled at length about the smart (pain) of modern mobile phones, but today I received support from an unexpected source:
I have always wished for my computer to be as easy to use as my telephone. My wish has come true because I can no longer figure out how to use my telephone.
The author? Bjarne Stroustrup, not exactly a beginner.
|
Walking round the Eagle's Nest
|
Topic: animals, general | Link here |
It's been a while since we took the dogs for a walk anywhere except in front of the house. Today we finally did it, taking them to the eagle's nest. It had been so long since they last travelled in the car that we had to show them how to get in.
Somehow there are far fewer trees in the area, not just because they fall down:
|
||||||||||||||||||||||||
|
Converting document formats
|
Topic: technology | Link here |
I've been asked to write an article about my experience with FreeBSD in the last third of a century. Really? That long? No, FreeBSD isn't quite 30 years old, but I've been using Unix on a desktop for nearly 33 years, and 31 of those were with BSD. Still, the part that got me was:
>> The process is fairly simple: you write the article (plain-text or Word
>> is fine)
>
> Now *that*'s a difference. Word! You mean Microsoft "Word",
> presumably. We wouldn't have even considered it in those days. I
> still wouldn't.
Hehe, the name certainly has established itself. Any kind of Word
processor (like Libreoffice) is fine, too.
How times have changed! I think I'll address this in the article. Should I submit in troff?
In that connection, though, it's worth saving this tidbit from the TUHS mailing list:
fwiw, Pandoc (https://pandoc.org) claims to be able to translate between MediaWiki and both man and ms roff macros.
|
MediathekView? MediathekWeb? MediathekViewWeb?
|
Topic: technology, multimedia, opinion | Link here |
Why have I suddenly had so many problems with MediathekView? Is it the old version I'm using? I don't think so, but it's worth checking. There's a web version available. www.mediathekweb.de? Maybe it was, but it's now in the hands of a domain squatter.
More searching: ah, isn't it obvious? www.mediathekviewweb.de? And yes, it has the same restrictions, though it can also download at even lower resolutions, in this case 291 MB instead of the 580 MB that MediathekView offers. I can't confirm, though: the page is terminally broken. I can't even decide if it thinks it can download or not. Still, it seems that whatever the problem is, it's not here.
| Monday, 30 January 2023 | Dereel | Images for 30 January 2023 |
| Top of page | ||
| previous day | ||
| next day | ||
| last day |
|
Hitler: the first 100 days
|
Topic: history, multimedia, opinion | Link here |
90 years ago today the German Reichstag made a serious mistake: they appointed Adolf Hitler as Reichskanzler. What happened then is history, of course. The ARD has brought out a TV series entitled „Hitler - Die ersten 100 Tage“ (“Hitler: the first 100 days”), and I'm currently watching it.
I've known about this for years—after all, it's basic history—and in fact the series isn't particularly good. It's depicted mainly from the diaries and opinions of about 6 people spread around Germany as it was at the time, including what are now Wrocław and Kaliningrad. But the way things happened is genuinely horrifying, and was to many people at the time. How could the government approve the Ermächtigungsgesetz, especially after the behaviour that Hitler's troops had already demonstrated?
In the past people have compared Donald Trump with Hitler, and I have disagreed. Trump was a total failure in every way. As this series emphasizes, Hitler wasn't, more's the pity. But the story should serve as a warning to modern people.
|
Insomniac distress
|
Topic: technology, opinion | Link here |
I've already grumbled about the fact that I can't put distress, my Microsoft “Windows” 10 box, to sleep. It keeps waking up. OK, one good thing about Microsoft is that you can probably find a solution to a problem online. Off looking, and came up with this page, which confirmed my suspicion that the problem was probably something in the system that was waking it again, and pointed me to a different way of the maze of twisty little passages that is Microsoft configuration: disallow wake timers.
Only: mine refuses to allow any change. Is that because it's on a remote desktop? It shouldn't make any difference, but who knows. On the other hand, I did find a way to get it to hibernate, so things are now marginally better.
| Tuesday, 31 January 2023 | Dereel | Images for 31 January 2023 |
| Top of page | ||
| previous day |
|
Dry omelettes revisited
|
Topic: food and drink, opinion | Link here |
Over the past year and a bit I've tried various ways to cook thin omelettes (intended for subsequent cutting into strips) without oil. The first attempt was in fact quite successful:
|
|||||||||||||||||||||
|
|||||||||||||||||||||
But there was a basic problem: the base of the pan is domed, and there's a danger that the middle of the “omelette” will get too thin and crispy before the outside is cooked, so I tried a “halo” with a hole in the middle, with little success:
|
|||||||||||||||||||||||
Why bother? What's a little oil between friends? Today I tried a smaller, flatter pan with some oil:
|
|||||||||||||||||||||||
|
|||||||||||||||||||||||
Somehow we're still not there. Maybe I need to experiment with halos.
|
Finding X fonts
|
Topic: technology, opinion | Link here |
One of the things that is definitely a problem with hydra are the X fonts, as I noted two weeks ago. How do I get a complete installation? Microsoft and Apple installations come complete; why not FreeBSD?
Off looking, and once again came up with a maze of little twisty possibilities, all different. One of the more interesting ones was to copy the Apple or Microsoft fonts. But there must be a simpler way.
|
FreeBSD network performance
|
Topic: technology, opinion | Link here |
FreeBSD has almost the reference implementation of the Internet protocols, but that doesn't mean that it's perfect. Currently there's a mail thread on the freebsd-stable mailing list, in which the original poster only gets a fraction of the throughput on FreeBSD that he gets on Linux.
Various suggestions on what to do, including setting undocumented sysctls—something that I have grumbled about in the past. I don't have time to look at the suggestions at the moment. One page that looks interesting is https://fasterdata.es.net/host-tuning/freebsd
Do you have a comment about something I have written? This is a diary, not a “blog”, and there is deliberately no provision for directly adding comments. It's also not a vehicle for third-party content. But I welcome feedback and try to reply to all messages I receive. See the diary overview for more details. If you do send me a message relating to something I have written, please indicate whether you'd prefer me not to mention your name. Otherwise I'll assume that it's OK to do so.
| Top of page | Previous month | Greg's home page | Today's diary entry | Next month | Greg's photos | Copyright information |
