animals brewing
food and drink fiction
gardening general
health history
language music
multimedia opinion
photography politics
Stones Road house technology
Greg's diary
recent entries
Translate this page
Select day in April 2020:
Su Mo Tu We Th Fr Sa
1 2 3 4
5 6 7 8 9 10 11
12 13 14 15 16 17 18
19 20 21 22 23 24 25
26 27 28 29 30
Select month:
2019 May Jun Jul Aug
2019 Sep Oct Nov Dec
2020 Jan Feb Mar Apr
Today's diary entry
Diary index
About this diary
Greg's home page
Greg's photos
Network link stats
Greg's other links
Copyright information

Tuesday, 7 April 2020 Dereel Images for 7 April 2020
Top of page
next day
last day

Updating eureka, attempt 15
Topic: technology, opinion Link here

I last updated eureka, my main machine, well over 4 years ago:

FreeBSD 10.2-STABLE FreeBSD 10.2-STABLE #2 r290972: Wed Nov 25 11:38:38 AEDT 2015  amd64

I've had a number of attempts to update it both before and after then, but there are a number of issues. Firstly, I hate to reboot. Now that we have the photovoltaic electricity installed, we no longer have power outages, and eureka has been up for 7 months, with promise of staying up much longer.

Then my X configuration is bizarre, and after 30 years of experience, X is still not as reliable as I would like. Upgrading my configuration to a new system has almost always given me issues, including last time.

Given the rarity of the event, should I perform hardware upgrades? eureka's motherboard is over 6 years old. If you believe Moore's law, modern processors should be 8 times as fast as then. Or is that 16 times? Depends on the law. But in fact there isn't that much difference. Yes, the AMD Ryzen Threadripper 3990X is the current top runner with 78,319 CPU marks (but only 2,607 per thread), compared to 6,881 for my Intel i7-4771 (which, I'm sure, was once rated only slightly below 10,000), but it wasn't the fastest in those days, just one that I considered affordable. The Threadripper costs $4,000, far more than what I would want to pay. And surprisingly the single thread rating of the Intel is 2,206, barely any different from the Threadripper. The fastest CPU under $400 US is currently another Threadripper, the AMD Ryzen Threadripper 1920X, rated at 24,489 CPU marks, not really worth the trouble. So I'll stick with the current processor and motherboard.

And what about disks? When I bought the current eureka, I included a 120 GB SSD for the root file system. There was no obvious difference in performance, and it died far too early. But clearly SSDs are the way of the future, and the prices have dropped significantly. It looks as if a 1 TB SSD would be a good idea: that would allow me to put just about everything I access frequently on the disk.

But looking at what's available, it seems that SATA SSDs (what I had last time) are a bad idea: they're limited to about 500 MB/s by the interface. The alternatives of PCIe and M.2, both of which can offer up to 4 GB/s.

M.2 requires dedicated slots on the motherboard, so that would require a different motherboard. But PCIe seems OK, and I have spare slots on the motherboard. But it's not clear that there's enough space between them and the graphics cards:
Image title: eureka 1          Dimensions:          4286 x 3635, 2883 kB
Make a single page with this image Hide this image
Make this image a thumbnail Make thumbnails of all images on this page
Make this image small again Display small version of all images on this page
All images taken on Wednesday, 8 April 2020, thumbnails          All images taken on Wednesday, 8 April 2020, small
Diary entry for Wednesday, 8 April 2020 Complete exposure details


That would require trying things out and being prepared to fail.

The real issue, though, is: how do I manage the upgrade so that I can continue to work even if something goes wrong? That's the real question. In the past, from 1992 until 2008, I ran two separate machines, joined together with x2x. That had the advantage that I never needed to take both systems down.

And maybe that's what I should be doing now. Many steps: split into two systems (in the process reviving dereel), upgrade dereel, upgrade eureka when dereel is working well, and at some later time put a 1 TB fast SSD in eureka. Motherboard upgrades postponed to some future date.

So today I found an old system (probably the old eureka) with an AMD Phenom 9550 and 6 GB of memory, and looked around for some spare disks with enough capacity. I have plenty of old disks, but most of them are small:
Image title: old disks 2
Complete exposure details
Dimensions: 771 x 350, 96 kB
Dimensions of original: 4583 x 2081, 1699 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Wednesday, 8 April 2020:
thumbnails    small images    diary entry

To my surprise, I found 2 1 TB disks (one marked “flaky”), a 2 TB disk, a 3 TB disk and even a 6 TB disk, the last out of teevee only 10 months ago. It'll make a good second disk, which I can mount on eureka or dereel.

For today I took the good 1 TB disk, which proved to have come from lagoon, and booted it as eureso:

=== root@eureso (/dev/pts/1) ~ 55 -> df
Filesystem     1048576-blocks      Used     Avail Capacity  Mounted on
/dev/ada0p2            99,176    35,625    55,616    39%    /
/dev/ada0p4            99,176         0    91,242     0%    /destdir
/dev/ada0p5           705,709   180,997   468,255    28%    /home

OK, first copy the root file system from eureka to eureso. While that was running, though, decided that it was high time to start again from scratch. I've been updating the system on eureka for at least 13 years, and some parts possibly as long as 30 years (starting with allegro in May 1990). I've already established that there are some strangenesses in eureka's mail configuration that I can't work out: after removing all traces of some aliases from the configuration, the server still accepted mail for them. I can't even guess where the information is stored. So starting from scratch would fix this and potentially a number of other issues.

Wednesday, 8 April 2020 Dereel Images for 8 April 2020
Top of page
previous day

Dual server install, next step
Topic: technology, opinion Link here

As planned yesterday, continued with my installation of a second system. The machine was already running an old instance of lagoon, masquerading as eureso, and it had a complete copy of eureka's root file system on another partition. But that wasn't what I wanted: I wanted a clean new install.

How do you do that with FreeBSD? I used to give classes in this sort of thing, but they started with a boot floppy disk, something mercifully obsolete. Nowadays we have bootable DVD images and bootable memory sticks. But that's not what I want; I just wanted a bootable root file system, and there seems to be no way to download one.

So I downloaded a DVD image and created a new virtual machine, then copied the file system to my new machine.

Various things happened. My first attempt to reboot failed: instead of rebooting, it hung in the “beep” that precedes rebooting. Hardware issues? It wasn't made any easier by the fact that I had connected the display to my eureka:0.0 monitor, which switches back to a different input when a display goes away. So I have no idea.

Made some process, in the process coming to the realization that I had migrated to real hardware (especially if it's flaky) far too early: I don't really need to take that step until almost everything is installed. And when I looked later on and discovered that the system had hung again, I decided that it was time to give up. Tomorrow I continue with a VM, where I can make a number of clean snapshots.

teevee crash!
Topic: technology, multimedia Link here

As if that wasn't enough, when I went into the lounge room to watch the news on teevee, it hung. Why? I couldn't get the firefox instance to stop. Into the office, where I could access teevee with no problems. The firefox process was stuck in a STOP state, something that I usually associate with a process dumping core. But I had disabled core dumps on teevee, and it was maxing out a processor core.

Dammit, restart the X server. But that was hanging too, waiting on some obscure kernel lock whose name I forgot to write down. A look at /var/log/messages showed:

Apr  8 17:04:05 teevee kernel: NVRM: GPU at PCI:0000:01:00: GPU-85983119-e9ef-ac66-6817-fbed5657b871
Apr  8 17:04:05 teevee kernel: NVRM: Xid (PCI:0000:01:00): 56, CMDre 00000001 00000094 00010049 00000007 00000000
Apr  8 17:04:05 teevee kernel: NVRM: Xid (PCI:0000:01:00): 56, CMDre 00000001 000000c0 0001004b 00000007 00000000
Apr  8 17:04:18 teevee kernel: NVRM: Xid (PCI:0000:01:00): 56, CMDre 00000000 0000045c 0001002c 00000007 00000000
Apr  8 17:04:18 teevee kernel: NVRM: Xid (PCI:0000:01:00): 56, CMDre 00000000 00000088 0001000a 00000007 00000000

Hardware problems? Probably. I wonder what it really means. But I just wanted to watch the news. Reboot the machine.

Oh. I still hadn't fixed the X configuration. OK, take a quick look, and decide that maybe it will work anyway. So I pressed the Big Red Button (black, power on/off) and rebooted.

Once again the automatic start of X failed, simply with the message that the server had stopped. OK, this time save the old log file to compare it with the new one. Start X manually. No problems. Remap the keyboard with the new, modified .xmodmap file. No problems.

So why didn't it start automatically? Off to take a look at the log files:

=== grog@teevee (/dev/pts/6) ~ 22 -> l -rt /var/log/Xorg*0408*
-rw-r--r--  1 root  wheel  43,858  8 Apr 17:03 /var/log/Xorg.0.log.old-20200408
-rw-r--r--  1 root  wheel  21,144  8 Apr 18:51 /var/log/Xorg.0.log-20200408

Oh. The old log file was the file from before the crash. What happened to the log file from the aborted automatic start? It seems that it didn't create one.

What is going on here?

In passing, took a look at the old log file to see whether it could help understand the hang. But no, there was just a disconnect/reconnect sequence, something that I have seen before, 777017 seconds (about 9 days) after the server was started. No idea how that relates to when the hang happened. Why does't X use clock times?

Topic: photography, general, opinion Link here

This morning Jari Kirma posted a photo on IRC:

Nice photo. It wasn't until some hours later that I discovered that it was a supermoon, apparently the biggest for who knows how long.

That photo looks as if it was taken with a mobile phone; the moon is clear, but it doesn't exactly dominate the image. OK, in the evening, for some reason, the sky was relatively clear, and I was able to get some photos with a longer focus lens:
Image title: Supermoon 2
Complete exposure details
Dimensions: 519 x 520, 153 kB
Dimensions of original: 1156 x 1159, 710 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Wednesday, 8 April 2020:
thumbnails    small images    diary entry
Image title: Supermoon 6
Complete exposure details
Dimensions: 600 x 450, 63 kB
Dimensions of original: 4608 x 3456, 2556 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Wednesday, 8 April 2020:
thumbnails    small images    diary entry
Image title: Supermoon 10
Complete exposure details
Dimensions: 600 x 450, 177 kB
Dimensions of original: 4608 x 3456, 6093 kB
Display this image:
thumbnail    hidden   alone on page
Display all images on this page as:
thumbnails    this size
Show for Wednesday, 8 April 2020:
thumbnails    small images    diary entry

But why is there chromatic aberration on the first image? It was centered, so normally any chromatic aberration would be the same on both sides, but here it's red on the bottom and blue on top. The moon was rising, but the shutter speed was 1/400 s, surely short enough to avoid any motion artefacts, and in any case, why should motion of that nature cause chromatic aberration?

The third photo shows a different problem: lack of dynamic range. It was exposed 11 EV more than the first image,but the trees in the foreground are still just silhouettes, and the moon is (obviously) overexposed. I should try some exposure bracketing if the weather is still acceptable tomorrow evening.

In passing, it's interesting that I can't detect any camera shake in the last image. The leaves are a particularly difficult subject in that regard. It was taken at 1/6 s with a focal length of 100 mm, which in this case corresponds to 200 mm “full frame”. My old rule of thumb told me that I would get camera shake below 1/200 s. Isn't image stabilization wonderful?

This page contains (roughly) yesterday's and today's entries. I have a horror of reverse chronological documents, so all my diary entries are chronological. This page normally contains the last two days, but if I fall behind it may contain more. You can find older entries in the archive. Note that I often update a diary entry a day or two after I write it.     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.

Greg's home page This month Greg's photos Greg's links

RSS 2.0 Valid XHTML 1.0!