#archlinux32 | Logs for 2018-07-15

[02:00:53] -!- City-busz has quit [Quit: http://quassel-irc.org - Csevegjen kényelmesen. Bárhol.]
[02:00:53] -!- elibrokeit has quit [Quit: A random quit message]
[02:02:23] -!- ztrawhcse has joined #archlinux32
[03:28:21] * buildmaster failed to execute a mysql query - can you have a look at "tmp.mysql-functions.query.stdin.2018-07-15T03:18:44.xX7RSS"?.
[04:35:28] -!- belanthor has quit [Quit: Leaving]
[04:45:43] -!- isacdaavid has quit [Quit: Leaving.]
[05:17:49] -!- davor_ has joined #archlinux32
[05:18:04] -!- davor has quit [Ping timeout: 260 seconds]
[05:18:04] davor_ is now known as davor
[05:20:26] ztrawhcse is now known as elibrokeit
[05:53:47] -!- muesli has joined #archlinux32
[05:53:56] -!- muesli has parted #archlinux32
[08:50:07] -!- deep42thought has joined #archlinux32
[08:50:08] <buildmaster> Hi deep42thought!
[08:50:18] <deep42thought> tyzoid: yeah, znc looks good :-)
[08:58:44] * buildmaster resumes sanity.
[09:21:33] -!- abaumann has joined #archlinux32
[09:21:34] <buildmaster> Hi abaumann!
[09:21:37] <abaumann> Hi!
[09:21:43] <deep42thought> Hi abaumann!
[09:21:45] <abaumann> Anybody experiencing mount issues?
[09:21:54] <deep42thought> not me
[09:22:07] <abaumann> My eeepc refuses to recognize the SD card at boot time and doens't mount it.
[09:22:18] <abaumann> the physical device is there, just systemd doesn't mount it.
[09:22:36] <deep42thought> I usually don't let systemd handle my mounts ...
[09:22:37] <abaumann> Now on my server (64-bit) similar issue with a two disk setup, UUID-mounted disks
[09:22:41] <abaumann> refuse to get mounted.
[09:22:53] <abaumann> systemd always handles your mounts..
[09:22:59] <abaumann> systemd-fstab-generator
[09:23:07] <deep42thought> no, it does not if you don't tell /etc/fstab
[09:23:44] <deep42thought> ok, I have two systemd-handled mounts - which work
[09:23:55] <deep42thought> and two not-systemd-handled ones, which work, too
[09:25:20] <abaumann> ok. I'm going through some reboot cycles then with debug enabled, etc.. :-)
[09:52:53] -!- City-busz has joined #archlinux32
[09:53:06] <abaumann> I get from booting into the ramdisk without error, only half of the disks mounted and all mounted any behaviour, setting debug for systemd mounts the disks, so I bet, it's a race in systemd.
[09:53:31] <deep42thought> does it show dependency warnings?
[09:53:37] <deep42thought> e.g. cyclic dependencies?
[09:54:09] <deep42thought> systemd has the tendency to ignore dependency errors (instead of failing totally) and thus allowing for races
[09:54:21] <deep42thought> ^ probably a feature
[09:55:43] <abaumann> why should cyclic dependencies work in debug mode only?
[09:56:00] <deep42thought> no, they shouldn't
[09:56:05] <deep42thought> but it may explain the race
[09:57:33] <deep42thought> my experience with systemd and mounts is just, that it tends to have cyclic dependencies and creates warnings + race conditions
[09:57:43] <deep42thought> that's, why I don't let systemd handle some of my mounts
[10:02:46] -!- eduardoeae has quit [Ping timeout: 264 seconds]
[10:26:20] <elibrokeit> deep42thought: how goeth python2 failures
[10:27:12] <deep42thought> elibrokeit: haven't looked into it much
[10:27:25] <deep42thought> the assertion is ok if CFLAGS are changed
[10:27:27] <elibrokeit> We're having fun too, don't worry.
[10:27:29] <elibrokeit> https://github.com
[10:27:31] <deep42thought> (have to find the minimal set, though)
[10:27:31] <phrik> Title:Testsuite fails on python 3.7 ¡ Issue #593 ¡ PyCQA/astroid ¡ GitHub (at github.com)
[10:28:19] <elibrokeit> note: this works fine on their interpreter, but some fairly minimal imports of their code lead to typeerrors
[10:32:24] <deep42thought> uh, btw: spi wishes to have clarified under what conditions we may use the name "Arch Linux 32" - e.g. how far we're allowed to deviate from archlinux
[11:03:36] -!- bill-auger_ has quit [Quit: No Ping reply in 60 seconds.]
[11:03:55] -!- bill-auger has joined #archlinux32
[11:07:07] -!- bill-auger has quit [Client Quit]
[11:07:56] -!- bill-auger has joined #archlinux32
[11:13:24] <buildmaster> firefox-developer-edition is broken (says rechenknecht).
[11:56:05] -!- abaumann has quit [Quit: leaving]
[12:28:18] -!- abaumann has joined #archlinux32
[12:28:18] <buildmaster> Hi abaumann!
[12:29:15] <abaumann> so, I downgraded systemd, same effect. I use genfstab -U / > /etc/fstab to create a vanilla fstab, same effect.
[12:29:19] <abaumann> now I see: fsck: Warning... fsck.ext4 for device /dev/sda1 exited with signal 15.
[12:30:32] <abaumann> so, error handling is just really bad.. :-)
[12:30:40] <deep42thought> :-)
[12:30:46] <deep42thought> what's signal 15?
[12:31:51] <abaumann> 15=8+4+2+1, so: Filesystem errors corrected System should be rebooted, Filesystem errors left uncorrected, Operational error
[12:31:54] <abaumann> ?
[12:32:15] <abaumann> but again. no mount.
[12:32:35] <abaumann> fsck executed manually shows it's clean..
[12:32:52] <deep42thought> maybe it's not yet available, when fsck is run by systemd?
[12:33:03] <deep42thought> whatever that^ means ...
[12:33:28] <abaumann> I seem to have a talent to hit strange bugs. :-)
[12:33:32] <deep42thought> shouldn't you be able to "systemctl restart" the failed units?
[12:34:09] <abaumann> I have the strange idea, that systemd is always in charge for the mounts, as it converts the /etc/fstab entries on boot to services.
[12:34:41] <abaumann> Inode 109576194 extent tree (at level 2) could be narrower. Fix<y>? yes
[12:34:49] <deep42thought> it is in charge, as soon as you create an entry in /etc/fstab
[12:34:52] <abaumann> aha. a forced check reveals errors on the partition.
[12:35:15] <abaumann> you don't mount the with a manual mount in a service, do you?
[12:35:16] <deep42thought> but it does not (yet) guess your mounts which do not appear in /etc/fstab :-)
[12:35:41] <deep42thought> I do mount manually all those, which systemd is unable to create the correct dependencies for
[12:35:53] <deep42thought> I have a somewhat strange setup, too
[12:35:56] <abaumann> It used to work a week ago, no suddenly it doesn't..
[12:36:10] <deep42thought> my /home is on an encrypted volume, whose key is accessed via sshfs
[12:36:15] <abaumann> but that's like using netctl instead of resolved, networkd (which failed also in the past).
[12:36:28] <abaumann> ok. this is a strange setup.. :-)
[12:36:35] <abaumann> mine is a SATA disk.
[12:36:56] <deep42thought> well, it works, if you let /home be mounted _after_ reomte-fs
[12:36:59] <deep42thought> which systemd does not
[12:37:59] <abaumann> what makes me really ponder is, that systemd handles some of the stuff available before, I can understand why they don't want to reimplement all strangeness. But it's not clearly stated, what IS implemented, and IMHO it's badly tested.
[12:38:35] <abaumann> * abaumann peeks at the console of the server during reboot..
[12:40:23] <abaumann> ok. the fsck exit code theory is also off the table.. again no partition mount
[12:40:38] <abaumann> 'mount -a' works.
[12:41:11] <deep42thought> does "systemctl restart $failedmount.mount" work?
[12:41:21] <deep42thought> (before "mount -a", obviously)
[12:41:31] <abaumann> * abaumann reboots :-)
[12:43:26] <abaumann> systemd-disk-by-id-uuid-xxxx shows a failed status.
[12:44:13] <abaumann> Jul 15 12:42:18 eurobuild3 systemd[1]: systemd-fsck@dev-disk-by\x2duuid-537b6cc8\x2d07e0\x2d4f67\x2d9f27\x2d3a24c1ef0cb0.service: Main process exited, code=killed, status=15/TERM
[12:44:17] <abaumann> Jul 15 12:42:18 eurobuild3 systemd[1]: Requested transaction contradicts existing jobs: Resource deadlock avoided
[12:44:20] <abaumann> Jul 15 12:42:18 eurobuild3 systemd-fsck[283]: fsck: Warning... fsck.ext4 for device /dev/sda1 exited with signal 15.
[12:44:23] <abaumann> Jul 15 12:42:18 eurobuild3 systemd[1]: systemd-fsck@dev-disk-by\x2duuid-537b6cc8\x2d07e0\x2d4f67\x2d9f27\x2d3a24c1ef0cb0.service: Failed with result 'signal'.
[12:44:26] <abaumann> so. manually fsck works, fsck via systemd in the ramdisk not.
[12:44:45] <deep42thought> this output is when you try to restart the unit manually?
[12:44:54] <abaumann> no. the unit doesn't exist.
[12:45:04] <abaumann> most likely because fsck sees an error.
[12:45:42] <abaumann> so , I can not restart the failed unit as the unit doesn't exist.
[12:46:06] <abaumann> ok. let's do a manual forces fsck again.
[12:46:39] <deep42thought> but some unit starts fsck, right?
[12:46:39] <abaumann> *forced
[12:47:01] <abaumann> systemd-fsck@dev-disk-by\x2duuid-537b6cc8\x2d07e0\x2d4f67\x2d9f27\x2d3a24c1ef0cb0.service loaded failed failed
[12:47:17] <abaumann> yes. that one is on failed, so I can rexecute that one.
[12:47:32] <abaumann> * abaumann twiddles his thumbs while this really big disk is being checked.
[12:48:01] <abaumann> Unit systemd-fsck@dev-disk-byx2duuid-537b6cc8x2d07e0x2d4f67x2d9f27x2d3a24c1ef0cb0.service has failed
[12:48:16] <abaumann> Dependency failed for File System Check on /dev/disk/byx2duuid/537b6cc8x2d07e0x2d4f67x2d9f27x2d3a24c1ef0cb0.
[12:48:41] <abaumann> Sorry.. this is overengineed crap..
[12:48:48] <abaumann> *engineered
[12:49:30] <deep42thought> :-)
[12:50:10] <abaumann> systemctl start systemd-fsck@dev-disk-by\x2duuid-062d61ac\x2dad44\x2d496b\x2db531\x2da83066b39df2.service
[12:50:13] <abaumann> hangs
[12:50:27] <abaumann> the other one is in a cyclic dependency with it's own fsck.
[12:50:51] <abaumann> and the messages of fsck end up where? I would expect in the journal. But nope, just 'error 15'
[12:52:19] <abaumann> let my try a forced check at boot time and some debugging enabled in systemd..
[12:53:04] <deep42thought> huh? where does that cyclic dependency come from?
[12:53:15] <abaumann> No clue..
[12:54:30] <abaumann> on my eeepc, the root device is mounted read-only/.
[12:54:59] <deep42thought> non-standard => breaks systemd
[12:55:15] <abaumann> it should be mounted read-only.. :-
[12:55:24] <deep42thought> yeah, that's what I meant
[12:55:26] <abaumann> the remount-rw failed in systemd.
[12:55:37] <deep42thought> you use a non-standard setup - why do you expect systemd to handle this?
[12:55:50] <abaumann> a non-standard setup?
[12:55:58] <deep42thought> "(12:55:15) abaumann: it should be mounted read-only.. :-"
[12:56:06] <abaumann> aeh...
[12:56:12] <abaumann> I meant: "read-write" :-)
[12:56:15] <deep42thought> ah, ok ...
[12:56:22] <deep42thought> then it _might_ be not your fault ;-)
[12:56:39] <abaumann> ..sorry. After 2 hours of trying to do basic stuff I start to get a little bit dizzy.. :-)
[12:56:51] <deep42thought> !grab abaumann
[12:56:52] <phrik> deep42thought: 🎉
[12:58:06] <abaumann> As I downgraded systemd with the same effects, I start to suspect a bug in e2fsprogs or in ramdisk.
[12:58:18] <abaumann> now I get again an emergency shell with no error messages..
[13:48:07] -!- abaumann has quit [Quit: leaving]
[14:31:21] -!- Alina-malina has quit [Ping timeout: 268 seconds]
[14:47:09] -!- baldemort has joined #archlinux32
[14:54:46] -!- Alina-malina has joined #archlinux32
[15:45:58] -!- abaumann has joined #archlinux32
[15:45:58] <buildmaster> Hi abaumann!
[15:47:48] <abaumann> so, I don't understand this modern stuff of mounting parallel mounts anymore. Disabling fsck in /etc/fstab seems to help (but is hardly a correct solution). I think, I apply the usual solution here, wait for the next systemd to fix it (and fuck up some other stuff)..
[15:48:06] -!- abaumann has quit [Client Quit]
[15:56:51] -!- buildmaster has quit [Ping timeout: 240 seconds]
[15:59:12] -!- buildmaster has joined #archlinux32
[15:59:45] -!- abaumann has joined #archlinux32
[15:59:46] <buildmaster> Hi abaumann!
[16:00:26] -!- abaumann has quit [Client Quit]
[16:11:18] -!- abaumann has joined #archlinux32
[16:11:18] <buildmaster> Hi abaumann!
[16:12:01] <abaumann> ok. I found nbd-server, nfs-server depend indirectly on bind mounts from a /data partition. This is a problem, as all things now run more in parallel (two physical disks get fsck and mounted in parallel).
[16:12:49] <abaumann> adding more and more threading and parallelity for making booting faster but the whole process harder to understand is failed engineering in my opinion.
[16:14:01] <abaumann> nicest thing: "unable to execute /bin/plymouth" when dropping to the emergency shell. I couldn't find a plymouth service.
[16:14:18] <abaumann> So, systemd is Redhadianising slowly slowly the Linux user land, not a good thing.
[16:15:29] <abaumann> * abaumann ends his "rant of the week" about systemd and watches soccer now. :-)
[16:15:30] -!- abaumann has quit [Client Quit]
[16:19:31] <buildmaster> libsavitar is broken (says eurobuild3).
[16:22:12] -!- eduardoeae has joined #archlinux32
[17:23:37] <deep42thought> abaumann: the core problem is messed up dependencies
[17:23:47] <deep42thought> once you solved that, systemd works as advertised
[17:42:18] -!- isacdaavid has joined #archlinux32
[19:46:55] <buildmaster> mame is broken (says tyzoid-srv0-bs0).
[20:01:16] -!- eduardoeae_ has quit [Quit: eduardoeae_]
[20:42:32] -!- belanthor has joined #archlinux32
[22:28:04] -!- deep42thought has quit [Quit: Leaving.]
[22:30:02] -!- deep42thought has joined #archlinux32
[22:30:02] <buildmaster> Hi deep42thought!
[23:48:28] -!- deep42thought has quit [Quit: Leaving.]