-
Notifications
You must be signed in to change notification settings - Fork 1.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Idle system, no obvious cause: PANIC at zfs_znode.c:620:zfs_znode_alloc(), VERIFY3(insert_inode_locked(ip) == 0) failed (-16 == 0) #12768
Comments
I believe the assert you hit was changed after so possibly still related. Not really sure why it was added at all tbh. |
Is the process 698277 (with the panic) trying to open one of the files that would have been affected by the rollback? During rollback the file inode(s) are released (via iput) asynchronously and when that is slow, we've seen variations on this problem. |
@PaulZ-98: pid 698277 is
Some research turns up |
This issue has been automatically marked as "stale" because it has not had any activity for a while. It will be closed in 90 days if no further activity occurs. Thank you for your contributions. |
This is the cause the crash.
Around 20:40 the same day I performed two rollbacks on two filesystems without unmounting them first to return to a clean state before restarting compiles jobs because the pool was full around 20:35 and some files were truncated. In the past, zfs rollback would hang occasionally on a filesystem without unmounting it first.
|
This issue seems to be similar to #11648 and #9741, but I chose to open a new issue because in my case there is no
recv
going on.System information
uname -a:
Linux $hostname 5.15.2-arch1-1 #1 SMP PREEMPT Fri, 12 Nov 2021 19:22:10 +0000 x86_64 GNU/Linux
ZFS-related packages installed via the
archzfs
repository:Describe the problem you're observing
This is my laptop workstation.
Chronology:
Symptoms
Diagnosis
txg_quiesce
in state D.zfs snapshot
process never finished, but was spawned after the hang, so it is not a cause but a consequence).sudo -u myuser strace zsh
and same forbash
show that both shells hang when attempting to interact with the history files in thehome
directory. The system root/
, the root user's homedir/root
, and the regular user's homedir/home/user
(under dataset/home
) are all in different datasets. Only/home/user
seems to be affected.ls -l /home
works fine.ls [-l] /home/user
hangs forever.ls -l /home/user/.ssh/
works fine ... !!!ls -l /home/.zfs/snapshot/autosnap_..../user
works fine./root
and copied to an external device. After rebooting the system, every file created under/root
after the crash was gone (including files and directories created at 15:10). I assumed the state of the rebooted system is the state of the system in the txg before the crash, but the systemd journal (in/var/log
, another dataset of this pool) seems to be complete with information up until 15:30. I have no idea what to take from this.zfs send pool/hostname/DATA/home@autosnap_....... > /media/external/file_home
worked fine, and I was able to keep a copy of the most recent snapshot of this dataset.zfs send pool/hostname/DATA/root > /media/external/file_root
hanged. It was expected to fail with errortarget is busy; if a filesystem, it must not be mounted
, and it's not even the dataset with apparent issues, but it hanged nonetheless.Possibly relevant
send
,recv
or evensnapshot
operations were ongoing.zfs-linux
was updated yesterday to match the kernel 5.15.2:[2021-11-14T04:37:10+0000] [ALPM] upgraded zfs-linux (2.1.1_5.14.14.arch1.1-1 -> 2.1.1_5.15.2.arch1.1-1)
zfs rollback
was executed for the first time on this pool yesterday afternoon, on thehome
dataset (the same one that hanged today), and the system was not rebooted between that rollback and the crash:2021-11-14.16:18:44 zfs rollback -r pool/hostname/DATA/home@autosnap_2021-11-14_16:00:00_frequently
Describe how to reproduce the problem
This is the first time it happens. I don't know how to reproduce.
Include any warning/errors/backtraces from the system logs
dmesg
Pool features
zpool get all pool | grep feature | sort -k3
More information
More information is available on request. Glancing over it, I can't see anything special going on there, so I did not include it in this report, since it's hard to sanitize.
Available information includes
ps -axuf
,pacman -Qi
,pacman -Sl
, the contents of/proc/spl/kstat/zfs
, includingdbgmsg
, and the output of the following sysrq commands:show-backtrace-all-active-cpus(l) show-memory-usage(m) show-registers(p) show-all-timers(q) show-task-states(t) show-blocked-tasks(w) dump-ftrace-buffer(z)
.The text was updated successfully, but these errors were encountered: