tiwi problems?
While looking at a web page on tiwi in the afternoon, everything froze. Well, almost everything. The mouse cursor still ran, but I couldn't input anything. Off to hydra, which told me that the X server was running at 100%. Rogue web page? Quite possibly. Shot down firefox, and X went with it. Restart X, and things seemed to work, but though I had started it with -listen tcp, I wasn't able to start anything that would talk to tiwi:0, not even locally. Is that because I started X from a different system? But while watching TV later, the system paniced: May 19 18:55:22 tiwi kernel: Fatal trap 12: page fault while in kernel mode May 19 18:55:22 tiwi kernel: cpuid = 3; apic id = 06 May 19 18:55:22 tiwi kernel: fault virtual address = 0x0 May 19 18:55:22 ...
Building enblend 4.1.4
I can't run the enblend 4.1.4 binaries directly on hydra due to library hell. Sometimes I wonder if dynamically linked executables are still a good idea. But I have the source: build it. Use the port in the Ports Collection? That's too fragile. If it's not up to date, it won't work. So build it as the author intended. After tripping over the confusing installation instructions, I came to: Extract the source archive, of course. I chose to do so in /usr/ports/graphics/enblend-4.1.4/work, giving the tree /usr/ports/graphics/enblend-4.1.4/work/enblend-enfuse-4.1.4.
SMS backup, the down side
After yesterday's success with saving SMS messages, I set up SMS Backup & Restore to back up at frequent intervals, once an hour. Today I downloaded the results. 20 files, each 20 MB in size, and all almost identical. Only the backup date was different. Complete backups! It's bad enough to have a single file backup, but it seems that there is no provision for incremental backups. Clearly once an hour is far too frequent. For the moment I'll go back to once a day.
No One Is the Enshittifier of Their Own Story
No One Is the Enshittifier of Their Own Story Today for my podcast, I read No One Is the Enshittifier of Their Own Story , my latest Locus Magazine column, about the microfoundations of enshittification: Therein lies the tale. The same people, running the same companies, are all suddenly behaving very differently. They haven?t all... more
More house photos
Another round of house photos today, and once again the latest version of enblend made a mess of them. Time to install enblend 4.1.4 on hydra.
Saving SMSs
I've been trying in vain for years to find an app that reliably forwards SMSs to email. This recent increase in their use by people who should know better just makes the thing more important. I have a couple of apps installed on my phone in the vain hope that they might help, but most of them just want to back up, not forward. OK, if that's all there is, try SMS Backup & Restore. Google Drive? Dropbox? OneDrive? Or (strongly discouraged) on your phone? Sigh. Somehow these people are not on the same planet as I. Select ?this phone?, ignore alarm bells, and I was actually able to select a directory folder that I could access externally, though of course they changed the path name from /SMS to /primary/SMS.
SMS takes over the world
I didn't need to worry about forgetting yesterday's doctor's appointment: I got not one, but two SMS to remind me. Marginally acceptable, I suppose. What if I hadn't read the messages? Would they have called me? If so, that's really acceptable: it saves a lot of work on both sides. But why not an email? Ah, email is portable, so it must be bad. Then this morning I received more SMSs: I suppose that's acceptable, too, though really barely. Why can't they just send email?
NFS fun
I still have my home directory spread across hydra and eureka, and it's definitely not ideal. Today I wrote yesterday's diary entry on hydra and committed it. It didn't show on the same file system on eureka. I've experienced consistency issues with NFS in the past, but it's been a while, and I've been doing things like this for a long time. What has changed now?
More hydra net hangs
Last month I had issues with hydra hanging while processing photos with despise, a Microsoft VM. I suspected a network card issue at the time, and I stopped using despise for a while. But then I tried again, and so far there were no issues. Then today I reprocessed some old photos, a total of 159 images from ten years ago. And in the middle of running ?Perfectly Clear?, I got another dreaded net hang. Or was it? I had prepared for this eventuality with a vty logged in as root and with only local file systems in its path. Switched to it.
Weight graph: take a step back
Yesterday's graph of my weight was compelling, but somehow it's a little short. It started at the beginning of the year, but I've been recording my weight for over a year. How about some more data points? Spent a while typing in the values for the last 4 months of last year. And how about that, a completely different viewpoint. The first graph is the second half of the second graph: Oh. That's not at all what I expected, and it clearly shows the advantage of plotting graphs.
The Colors of Racism
Recently, somewhat by accident, I stumbled into reading a couple of monstrously racist texts, and I?m going to need to update the Wikipedia entry for a famous British author. But I learned a few things along the way that I want to share. Disclosure I try to be antiracist, but I don?t think I?m particularly good at it. I sometimes have bigoted feelings but try hard to recognize and not act on them. I?m convinced that humans are naturally tribal and antiracist work will continue to be required for the foreseeable future.
Recovering hydra, day 2
Copying the /home file system on hydra ran until 4:28 this morning. The next step would be to copy the individual files?maybe. After all, they were all backed up?maybe. This morning's backup output failed: *** Mail of at least 153911296 bytes could not be accepted *** at hydra.lemis.com due to lack of disk space for temp file. *** Currently, 150424 kilobytes are available for mail temp files Admire the different units, almost obscuring the fact that the message was only just too big. I wonder what it contained.
Catastrophe!
After yesterday's power outage, I left it until today to pick up the remains of hydra. A good thing, too. It took all day and I still wasn't done. In principle it came back relatively quickly. Fire up my kludge window managers, start firefox... Nothing. Well, for a while. Then the machine rebooted. Huh? That never happens. Try again. Start X. Start window managers. Start firefox. Spontaneous reboot! Damn, what's causing that? Something wrong in /usr/local? Ran a manual fsck on the root file system, and how about that, it found a number of anomalies. Start firefox. Crash. OK, something in /usr/local?
Power outage!
While watching the news on TV this afternoon, there was a ?beep? and the TV and tiwi, the driving computer, powered off. Nothing else, including the microwave ovens and lagoon, Yvonne's computer, which were on the same circuit, didn't. It must have been a power surge, lasting less than a second. Damn! I must do something about that. Waited for it to come back. It didn't. And of course the display remained blank. I really need to do something about that; the TV is connected by HDMI, but the machine comes up looking for the DVI interface, which wasn't connected. Into the office to get a monitor to connect to the DVI cable which I had hanging out for exactly this purpose.
ACDSee software free
Interesting offer in the mail today: ?Completely FREE! Get Gemstone Photo Editor now, no strings!?. OK, if it's free, I'll take 2. I looked at ACDSee briefly ten years ago, and while I didn't find anything wrong with it, I didn't mention it again. OK, follow the link. ?Please fill out this CAPTCHA?. Dammit, what is wrong with the browser on hydra:0.2 that always wants CAPTCHAs to be filled out? Started up despise, where it's supposed to go, and got the same result. Then a download link, along with a popup asking whether I wanted to do something that I could resist.
More Hugin experiments
Yesterday's comparisons of the output of enblend 4.1.4, 4.2 and 4.3 were revealing, but they don't match the images I used last week.
Hugin problems, continued
Last week I reported the problems I had with the new version of Hugin and got a response from Lukas Wirz, who is working on enblend and has recently committed a patch to fix this kind of problem. So I downloaded the latest version of enblend 4.3 from https://sourceforge.net/p/enblend/code/ci/default/tree/, a link that I had to be told about: the home page only offered version 4.2, which is what I had been running. Building had also not been straightforward: the FreeBSD Ports Collection has a too-rigid structure to easily deal with changes of location and archive format, and the README in the archive is seriously in need of correction.
More net outages
Another two National Broadband Network outages this morning. Clearly we can expect more: And reading back in my diary for years gone by, that's nothing unusual. Five years ago today I had exactly the same problem.
libmysqlclient: still no joy
The power outage also meant that I had to manually restart the weather monitor on tiwi. As last month, I had to run ldconfig to get the system to recognize it. So it wasn't the symlink. Presumably the issue was that MySQL wasn't installed on tiwi, and the hints file doesn't include the directory. Time to RTFM and check which hints file I should update.
Another bloody RCD trip!
While making breakfast this morning, the RCD for the kitchen tripped. That's the same one that has given me pain last month. And on that occasion, too, it happened while cooking breakfast. I had suspected the water kettle then, but there was no obvious connection today. Still, since it also took tiwi down, it might be time to replace the kettle.
Hacking our way to better team meetings
My team and I set out to build a simple note taking aide, which transcribes and summarizes our meetings using Bedrock. Today, we?re making the proof of concept available to everyone.
More network outages
Another network outage round 9:00 this morning, like last week. There was a difference, though: the display on the NTD was different, no signal strength indication at all. Clearly one of the blanket National Broadband Network outages that they're threatening all the time. OK, now that I can access the MyAussie app without too much pain, let's see what they have to say. Yes, of course we don't have an NBN connection, which is why I enabled mobile data. But even if I hadn't, ?something went wrong? is a cop-out. And despite everything, I wasn't able to access the site: Report a fault?
Rearranging the office
I've been in my current office for nearly 9 years, and in that time a lot has not changed. I still have the same computer and some of the monitors, and the computers under the monitors have also not all changed. Here the only photo I took 9 years ago, and how things look now: But I put a couple of audio amplifiers under the leftmost and rightmost monitors. The leftmost was my old Technics SU-3500, but it failed relatively soon after, so I replaced it with a Sansui San110, which ended up under the right-hand monitor.
More weather station pain
Writing yesterday's article on my weather program, I checked the current status. Nothing! It seems that it had stopped logging at 17:30 or so yesterday afternoon. Checking came up with a continual stream of May 5 07:00:03 tiwi wh1080[111]: Can't read device: Input/output error or Unknown error (5) May 5 07:00:03 tiwi wh1080[110]: Can't read device: Device busy or Unknown error (16) May 5 07:00:06 tiwi wh1080[116]: Can't read device: Device busy or Unknown error (16) May 5 07:00:06 tiwi wh1080[115]: Can't read device: Input/output error or Unknown error (5) May 5 07:00:08 tiwi wh1080[119]: Can't read device: Input/output error or Unknown error (5) What went wrong there?
More Hugin strangenesses
It was house photo day yesterday. And once again I had issues with the newer version of Hugin that I have on hydra (2023.0.0.d88dc56ded0e) while the older version on eureka (2018.0.0) worked fine. I use a script that creates the project files (.pto) and then stitches them. Yesterday's panoramas had a number of problems, the most extreme of which was this one (first 2023 Hugin, then 2018: What causes that? The project file is the same in each case, and so is the invocation to build the panorama, in essence hugin_executor --stitching laundry-door.pto Could it have something to do with masks?
Storage Churn
What are the highest-impact Cloud services? Storage would be near the top of any list. Where by ?Storage? I mean what S3 does: Blobs-of-bytes storage that is effectively unlimited in capacity, credibly more durable than anything you could build yourself, and easily connected to the world, either directly or through a CDN. I think we?re entering a period of churn where there?s going to be serious competition on storage price and performance. Which, by the way, is crucially relevant to the Fediverse. Let?s start with AWS, since they invented the modern Storage concept. The most important thing about S3 is this: There appear to be zero credible reports of S3 data loss.
Ramen for breakfast?
I've been thinking of various ways to eat Ramen noodles for breakfast. One of the more interesting facts is that they really come from China, and the kind I have come from Shandong. I've played around a couple of times, but how about asking an artificially intelligent expert? To my surprise, Google Gemini came up with a recipe that doesn't look bad. The interesting idea was to fry the noodles until crisp. That didn't work for me, and in general the experiment wasn't very successful. In particular, I didn't know that the Miso was salty, so the overall dish was just barely edible.
More ldconfig strangeness
Rebooting tiwi meant that my weather program didn't get restarted. Once again I had this issue with libmysql.so.18. I suspected that that was due to the fact that the directory usr/local/lib/mysql was a symlink to the directory on eureka. OK, copy the files over and run ldconfig on them. It still didn't work! I was in the parent directory, so I simply wrote ldconfig -m mysql. And it seems that it doesn't like that. ldconfig -m /usr/local/lib/mysql/ did the trick.
The night when time stood still
Woke up at 5:15 this morning. What's that light outside? It was almost as bright as day. But the sun doesn't rise for another 2 hours. Turned around, went back to sleep and woke at 7:15, but couldn't get back to sleep. OK, get up. Looked at my watch. 8:42! My bedside clock was fully 1 hour, 37 minutes slow! How could that happen? The clock is electronic and synchronizes to mains frequency (which is somewhat fast). It doesn't seem likely that it would fail like that. And then I asked Yvonne. Yes, her clock stopped altogether round 1:15. She looked back quite some time later and it was showing the same time, so she reset it.
Web site overload: recovery
Once again this morning the web server error log was full of error messages from ip7.ip-145-239-202.eu. But finally it seems to be over. All that remains is this strange error message [Thu May 02 07:42:26.457757 2024] [php7:notice] [pid 27107] [client 145.239.202.7:64277] PHP Notice: Undefined index: -1 OR 2+948-948-1=0+0+0+1 in /home/grog/www.lemis.com/grog/kleins-road/exterior-slides.php on line 399, referer: http://www.lemis.com/grog/diary-jun2013.php?subtitle=Al%20Jazeera%20issues&article=D-20130604-000628 OK, I've blown away that file, but I have a local copy. Works perfectly. Copy it back to www. No photos found! Potentially that was the reason for the negative index.
Android strangeness
Downloading some files from hirse, my mobile phone, was particularly slow today. Why? === grog@hydra (/dev/pts/15) ~ 4 -> ping hirse PING hirse.lemis.com (192.109.197.228): 56 data bytes 64 bytes from 192.109.197.228: icmp_seq=0 ttl=64 time=4432.006 ms 64 bytes from 192.109.197.228: icmp_seq=1 ttl=64 time=4491.037 ms 64 bytes from 192.109.197.228: icmp_seq=2 ttl=64 time=4636.822 ms 64 bytes from 192.109.197.228: icmp_seq=3 ttl=64 time=4777.693 ms 64 bytes from 192.109.197.228: icmp_seq=4 ttl=64 time=4885.086 ms 64 bytes from 192.109.197.228: icmp_seq=5 ttl=64 time=3869.501 ms 64 bytes from 192.109.197.228: icmp_seq=6 ttl=64 time=2954.141 ms 64 bytes from 192.109.197.228: icmp_seq=7 ttl=64 time=1959.271 ms 64 bytes from 192.109.197.228: icmp_seq=8 ttl=64 time=967.665 ms 64 bytes from 192.109.197.228: icmp_seq=9 ttl=64 time=4.943 ms 64 bytes from 192.109.197.228: icmp_seq=10 ttl=64 time=1.908 ms 64 bytes from 192.109.197.228: icmp_seq=11 ttl=64 time=1.314 ms 64 bytes from 192.109.197.228: icmp_seq=12 ttl=64 time=2.119 ms It came good by itself.
Web site attack
Sometime in the early evening it occurred to me that I hadn't received any external mail in over an hour. Problems on the external server? Yes. The file system was full. This morning there were something like 13 GB free, but now it was at 108% with -3.5 GB. Something had eaten up over 16 GB in a few hours. I knew where to look. The web server error log was full with messages like [Thu May 02 07:42:26.457757 2024] [php7:notice] [pid 27107] [client 145.239.202.7:64277] PHP Notice: Undefined index: -1 OR 2+948-948-1=0+0+0+1 in /home/grog/www.lemis.com/grog/kleins-road/exterior-slides.php on line 399, referer: http://www.lemis.com/grog/diary-jun2013.php?subtitle=Al%20Jazeera%20issues&article=D-20130604-000628 [Thu May 02 07:42:26.457785 2024] [php7:notice] [pid 27107] [client 145.239.202.7:64277] PHP Notice: Undefined index: -1 OR 2+948-948-1=0+0+0+1 in /home/grog/www.lemis.com/grog/kleins-road/exterior-slides.php on line 399, referer: http://www.lemis.com/grog/diary-jun2013.php?subtitle=Al%20Jazeera%20issues&article=D-20130604-000628 [Thu May 02 07:42:26.457812 2024] [php7:notice] [pid 27107] [client 145.239.202.7:64277] PHP Notice: Undefined index: -1 OR 2+948-948-1=0+0+0+1 in ...
www: no swap!
A few months back I ran into memory pressure on www.lemis.com, my remote virtual server. OK, add a ?memory disk?. And that worked. # dd if=/dev/zero of=/var/swapfile bs=1m count=4096 # mdconfig -a -t vnode -f /var/swapfile md0 # swapon /dev/md0 Until Vultr decided to reboot my machine. And then the md device was gone. It wasn't until today that I noticed. How do you ensure that they're recreated after reboot?
Cracking the fvwm2 to fvwm3 transition
I've been dragging my feet on updating hydra's window manager, mainly because I don't know where to start. I wrote the configuration files nearly 30 years ago, and I have long forgotten the details, in particular what I changed relative to the defaults: this was before I religiously ran version control. One of the big issues, though, was that fvwm3 no longer issues error messages! So I don't know what it doesn't like. After RTFM I discovered that I can get it to output messages to a log file, apparently with a predetermined name ~/. ~/.fvwm/fvwm3-output.log. That's particularly irritating in my situation, where I don't have a directory ~/.fvwm.
Academia nonsense of the day
After being asked if I wrote ?Treasurer? and co-authored a paper with PO Box, today I got another one: 21 N 01-05-2024 To groggyhimsel ( 901) Mentioned by rm rf N ?Greg Lehey? cited by ?rm rf? Is this Artificial Intelligence at its finest?
NBN outage!
Early this morning, while I was reading news on hirse, my mobile phone, an app failed to resolve a host name. Bloody Android apps! It wasn't until later that I discovered that that was the beginning of a 30 minute network outage. OK, what's my MyAussie app for? Fired it up and discovered that it wanted a password. My passwords don't cater for mobile phones: this one was 31 characters long, sprinkled with special characters that are a particular pain to enter on a mobile phone (is that why so many password restrictors demand them?) , and Aussie is too secure to display more than one character at a time when you enter a password.
NetBSD again
I have a bug report on makefs, a program ported to FreeBSD from NetBSD. It works under NetBSD, but the FreeBSD version creates invalid file systems under some circumstances. OK, let's compare what FreeBSD and NetBSD do. For that I need a NetBSD system, and the last one I had (a virtual machine) is ancient. OK, install a new virtual machine. How hard can it be? The biggest problem was getting installation media. For some reason, both firefox and fetch failed early in the download. In the end I downloaded it to my external server and then copied it from there.
Security, 2024 style
Mail from eBay today. I have a second account that I set up for some purpose, and they've noticed that I don't use it. Can they shut it down? No, I think not. I don't know what I might need it for, but it's convenient. All I need to do is log in. I failed. Password wrong. No worries: they'll send me a reset link. Got that, went to the login page. Enter user ID. OK. Enter new password. It's not strong enough! Must contain at least one digit and one special character to make it easier to crack. I tried, in sequence, No security!