https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=288345
--- Comment #17 from Sebastian Oswald <[email protected]> --- This week I had another 4 (!) poudriere bulk jobs hanging indefinitiely and was unable to terminate them and couldn't reboot without a power cycle via ipmi... For the last occurence, I could determine the approximate moment when all builds went catatonic and that time pretty much lined up with these cronjobs: 0 22 * * * /usr/local/sbin/zfsnap snapshot -a 3w -r nvme_pool/poudriere/data 0 22 * * * /usr/local/sbin/zfsnap snapshot -a 4w -r nvme_pool/poudriere/jails 0 22 * * * /usr/local/sbin/zfsnap snapshot -a 4w -r nvme_pool/poudriere/ports 0 22 * * * /usr/local/sbin/zfsnap snapshot -a 3m -r nvme_pool/poudriere/poudriere.d These were present as long as this system existed and always worked. I suspect it's the first one causing issues, as it's always the data/.m mounts that go titsup. The fact that the snapshot itself seems to trigger all builds to go catatonic and make the data/.m mounts defunct definitely reeks of some regression with one of the latest zfs updates (of which there are way too many nowadays IMHO...), so poudriere most likely isn't at fault here. -- You are receiving this mail because: You are the assignee for the bug.
