| systemdlete | We are good again... for some reason. I rebooted the VM numerous times, trying different configurations, and in that process, I noticed that fsck ran on some of the file systems. That may be it, and it would make some sense if that were the case: Right before the blank screen problem started, I could not even run ls or some other | 00:54 |
|---|---|---|
| systemdlete | programs; I kept getting "segmentation fault" -- so it could well have been a corrupted file system that needed an fsck. | 00:54 |
| systemdlete | Then again, why the fs corruption on Linux in 2022? Hmmm. | 00:54 |
| Hurgotron | systemdlete: The last time I had weird issues with segfaulting binaries and wrong checksums it was bad memory. If a bad buffer gets written back, that *may* also cause fs corruption. - Just an idea. | 01:00 |
| systemdlete | hmmmm. So maybe time for a memcheck86 ? | 01:00 |
| systemdlete | thanks for reminding me of that. | 01:00 |
| phogg | every time I've seen fs corruption it has been hardware failure | 01:00 |
| systemdlete | I hope not though. Finding replacement memory for this old board might be challenging. Just finding mainboards this old is hard enough! | 01:01 |
| systemdlete | I wonder, though. Could CPU starvation do something similar? | 01:01 |
| systemdlete | The VM seems to have plenty of memory for what little it does. | 01:01 |
| systemdlete | Just bumped it from 3 to 4 vCPUs today. | 01:02 |
| Hurgotron | systemdlete: I'm actually running ECC memory since then. It took memcheck *18 hours* to start finding issues, and I had issues for months before I finally ran memcheck that long. I don't want that ever again. | 01:02 |
| systemdlete | Yes. ECC memory. Idk why, but for years, I was told to avoid it for some reason or another. | 01:03 |
| systemdlete | I mean, probably 10-15 years ago. | 01:03 |
| systemdlete | (about the age of my hardware!) | 01:04 |
| systemdlete | I'm still running FX8350 and Thuban II X6 | 01:04 |
| systemdlete | DDR3 memory | 01:04 |
| Hurgotron | systemdlete: With Intel, your choice is rather limited for ECC, mostly server hardware. But most AMD CPUs support it, and most boards too (without even mentioning), except maybe a few very cheap ones. | 01:17 |
| gnarface | systemdlete: stand by, gotta read the scrollback | 01:19 |
| systemdlete | gnarface: Thanks. But I think it has kind of "resolved" itself. I am ordering ECC memory, partly for upgrade and partly so I can build another box. (I also have 2 mainboards coming, along with 2 FX8350s that are sitting here ready). | 01:21 |
| systemdlete | Still, some understanding of the issue would help settle my gut atm. This really threw me. My first thought was, OMG, I just updated all that garbage for polkit, and since I have had no particular issues recently with the OS, that must be the smoking gun. | 01:22 |
| systemdlete | Of course, it wasn't. Just some file system corruption. (e_e) | 01:24 |
| systemdlete | another possibility, of course, is virtualbox. It has caused some hiccups through the years, although overall, it has been quite good to me. I am looking at other solutions for virtualization, but so far, I prefer the vbox gui the best (and it is lacking in many ways too!) | 01:26 |
| systemdlete | bbl to read your responses. | 01:27 |
| gnarface | systemdlete: yea i would have been going crazy too. if you're sure now it's filesystem corruption, the one conspicuous non-hardware related filesystem corruption issue i've had recently i think was due to a known issue with using e2fsprogs from a specific older release on a newer version of the ext4 filesystem | 01:33 |
| gnarface | systemdlete: i forget which exact release actually now but i think it was something that i ran into upgrading from or to jessie | 01:35 |
| gnarface | systemdlete: any chance you might have fsck'd or resized a ext4 filesystem on one of these things using older versions of the tools, maybe during a release upgrade or some repair process? | 01:35 |
| Xenguy | Does anyone have info on that recent polkit bug? I'm still running Beowulf, and wondering if this has been patched? | 02:05 |
| gnarface | yea there were patches up in beowulf by the next morning | 02:05 |
| Xenguy | Thanks gnarface , I saw 2 or 3 updates go by, but wasn't paying enough attention to know what it was | 02:06 |
| xisop | i imagine the polkit patch is a line or two.. just returning early if argc is zero | 04:26 |
| * systemdlete returns to the convo, 10 wings and a beer later... | 04:30 | |
| systemdlete | gnarface: I suppose it is possible, but I'd think the error(s) would have shown itself sooner, no? | 04:30 |
| systemdlete | I mean, it could have become an embedded, latent bug, but I think the last time I would have touched it was months ago. | 04:31 |
| systemdlete | Oh... | 04:31 |
| systemdlete | you know what? You might be on the money, actually, gnarface. About a month ago or so, I accidentally dd'd an ISO to what I thought was the thumb drive, and hence came much drama and sweating and pulling my hair out. By some miracle/luck, I was able to re-create the fs and restore the files on the virtual drive from my backups (thanks | 04:35 |
| systemdlete | bareos!) and got things back to normal. Or at least I thought so. I have not seen major problems with this VM. | 04:35 |
| systemdlete | One thing I am not sure of, though, is if I might have used a linux rescue cd. But even if I did, I would hope their fs tools were up to date. I'd hate to think of anything *rescue* that did not have recent tools. | 04:35 |
| systemdlete | Whether it was restore, rescue cd, or a combo of both, somehow the system came back to life with nary a complaint from it. | 04:36 |
| systemdlete | What I did to my poor VM(!) is just awful and not something I am proud of. I have enough years of experience to know better than to blindly start writing to a raw disk drive, virtual or otherwise. However, I did what I could to recover, and thought I was OK. | 04:37 |
| systemdlete | Anytime I go to write what I think is a thumb drive in the future, I will be sure to be extra careful... | 04:38 |
| systemdlete | please excuse all my noise in the channel everyone. | 04:39 |
| gnarface | systemdlete: no no, it was for sure actually quite quiet, a very quiet error that silently unravels some part of the filesystem deep in some random subdirectory, and you don't notice until something tries to access it again, but then when something does... the corruption grows | 04:45 |
| gnarface | systemdlete: in that respect it behaved exactly like hardware failure - that was actually the false positive that threw me off for a while on the issue (blaming the SD cards) | 04:46 |
| systemdlete | In most respects vb "hardware" behaves pretty much like the real thing, with the exception perhaps of the networking, which I have noticed has a tendency to sort of "back up" and then release a bunch of packets at once. But that only happens occasionally, usually when I have poorly configured my network somehow (bad router config, wrong network | 04:48 |
| systemdlete | attached to a vbox interface, etc) | 04:48 |
| gnarface | systemdlete: if you tar from one filesystem to another, you might be able to strip the corruption out of a backup and go again | 04:48 |
| systemdlete | Probably best to reformat the partition then tar/untar | 04:49 |
| systemdlete | that's how I move existing partitions to new (usually larger) partitions | 04:49 |
| systemdlete | that is, if the partitions are on drive partitions rather than LVM, which I much prefer for just this reason. | 04:50 |
| systemdlete | and the partitions on this VM were LVM LV's | 04:50 |
| gnarface | you asked about other virtualization solutions, i'd just use bare qemu if you need hardware replication | 04:51 |
| systemdlete | the VGs span several physical volumes. | 04:51 |
| systemdlete | Forgive me, but I admit to enjoying a GUI | 04:51 |
| systemdlete | even if it is not the best it could be | 04:51 |
| systemdlete | I've looked at VMM and aqemu but both have issues of different sorts | 04:51 |
| gnarface | i don't really blame you about the GUI thing but i despaired of that a long time ago... | 04:52 |
| systemdlete | VMM creates a bridge device for every single interface (rather than just a tun device) | 04:52 |
| gnarface | are you using iptables? | 04:52 |
| systemdlete | aqemu looks nice but has god knows how many bugs. Plus the qemu command line it generates is for an older (prob c. 2016) qemu | 04:53 |
| systemdlete | I use ufw/gufw for firewalls. Plus I have several virtual routers running openwrt | 04:53 |
| gnarface | nftables frontend? | 04:54 |
| gnarface | hmmm | 04:54 |
| systemdlete | All of these tools, I openly admit, have their share of issues. But I still like being able to look at a problem with a high, visual view because it helps me reckon with at least superficial problems like wrong configs etc | 04:54 |
| systemdlete | not sure about nftables, but I think iptables has been re-implemented in nptables now, right? Or do I have that wrong? | 04:55 |
| systemdlete | I thought iptables had become just a frontend to nftables now. | 04:55 |
| gnarface | supposedly | 04:55 |
| gnarface | i don't know really | 04:55 |
| gnarface | i just wonder how optimized those rules it makes are | 04:55 |
| systemdlete | Oh, I only use ufw/gufw for my VMs. The routers have their own (LuCI) web front ends. | 04:56 |
| systemdlete | but... we should continue this in the chatter channel | 04:57 |
| systemdlete | thank you for your insights on this. They really made me re-think the whole thing. It started with a very careless mistake. | 04:57 |
| systemdlete | the feedback I've been getting in #devuan has been very friendly and constructive, unlike some other distro's IRC's | 04:58 |
| brocashelm | ^^^ agreed | 05:01 |
| blizzow | I installed the discord .deb and the thing works but broke apt. apt now complains that I need libappindicator1 and libc++1 but they seem unavailable in daedalus. Anyone got an idea how I can fix this without uninstalling discord and reinstalling it every time I do something with apt? | 06:18 |
| blizzow | Also, does anyone here have experience managing devuan hosts with ansible? | 06:22 |
| Afdal | I notice some polkit updates in chimaera | 06:26 |
| Afdal | but I don't see anything about pkexec | 06:26 |
| Afdal | Has Devuan managed to patch out this recent exploit yet? | 06:26 |
| fluffywolf | policykit-1 is indeed packaged by devuan, but I don't know if it's been patched yet. | 06:31 |
| bb|hcb | it is | 06:32 |
| bb|hcb | pkexec is part of policykit-1 | 06:33 |
| fluffywolf | looks like version was bumped yesterday, which suggests it might be. | 06:33 |
| fluffywolf | bbl, wolfy bedtime | 06:34 |
| fluffywolf | joerg will know! :P | 06:34 |
| blizzow | Is there a repo with non-free firmware available? I added non-free to my /etc/apt/sources.list entry ran apt update and still am unable to install non-free firmware. :( | 07:05 |
| argie_ | Hey blizzow. There are many non-free packages in the non-free area. Is it one specific package you're not able to find? | 07:28 |
| argie_ | There are many non-free firmware packages* | 07:29 |
| blizzow | argie_, just the iwlwifi firmware. | 07:32 |
| gnarface | it's called firmware-iwlwifi | 07:34 |
| gnarface | you can search by regexp | 07:34 |
| gnarface | it helps | 07:34 |
| gnarface | dunno about how to fix the discord thing, maybe install it to a chroot so it doesn't screw up your main install | 07:35 |
| joerg | hm? sorry reconnect, lost any context for that, fluffywolf | 08:49 |
| joerg | fluffywolf: hmmm? http://reisenweber.net/irclogs/libera/_devuan/_devuan.2022-01-28.log.html#t2022-01-28T06:34:36 | 08:51 |
| joerg | ok, you'll pay for those wasted 5 minutes | 08:54 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!