=== shokohsc517 is now known as shokohsc51 | ||
webchat8 | a RAID 5 mdadm array with a write journal has started segfaulting the kernel after upgrading 22.04 LTS to the latest 5.19.0-50 kernel. this crash happens within a couple hours, and results in unclean shutdowns of arrays, file systems, and sometimes hangs the rebooting process as well requiring pressing the reset button on the machine. | 20:03 |
---|---|---|
webchat8 | suspecting the write-through journal is at fault (because previously a write-back journal periodically hung the array and resulted in more unclean shutdowns), I tried changing the array back to use a bitmap consistency policy instead. I have not been able to disable the write journal and moving the array back to using a bitmap, or even to resync. | 20:03 |
webchat8 | mdadm's manual suggests doing that is not well supported, but circumstantial evidence says removing a write journal may be done by failing then removing the journal device from the raid 5 array. the procedure is not documented. do you have any suggestions on how to proceed here? | 20:03 |
sarnold | I think if it were me I'd try to reproduce the setup with VMs and test a bunch of the choices there, before doing anything on a filesystem that would be timeconsuming/expensive to restore from backups | 20:04 |
webchat8 | that's always a choice, the restore from backups route. fortunately it's not that much data. and of course multiple such backups exist. | 20:09 |
webchat8 | but what should I make of "mdadm segfaults the kernel"? | 20:09 |
sarnold | webchat8: it'd be nice to get a bug report for that one :) | 20:17 |
Yakov | having device /dev/ezcap -> video1, it seem to be busy, how to find processes related to the device? | 21:21 |
tomreyn | lsof | 21:24 |
Yakov | having device /dev/ezcap -> video1, it seem to be busy, how to find processes related to the device? | 21:31 |
oerheks | lsof /dev/video1 | 21:32 |
oerheks | or ezcap ? | 21:32 |
Yakov | it works for same namespaces, thank you, oerheks | 21:37 |
Yakov | having some ghost commands, know PID | 21:38 |
Yakov | how to kill? | 21:38 |
oerheks | ps aux | grep Z ## then "kill PID" where PID is the process ID of the zombies process | 21:40 |
oerheks | or if you know what application you started; pkill <application> | 21:41 |
Yakov | COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME | 21:42 |
Yakov | memcheck- 11082 supernova 6u CHR 81,1 0t0 848 /dev/video1 | 21:42 |
Yakov | cant kill it | 21:42 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!