Summary: | 12.0 Release: tmpfs (used by Postgres 12.0) unkillable hang | ||
---|---|---|---|
Product: | Base System | Reporter: | Viktor Dukhovni <ietf-dane> |
Component: | kern | Assignee: | freebsd-fs (Nobody) <fs> |
Status: | Closed Overcome By Events | ||
Severity: | Affects Only Me | CC: | chris, markj |
Priority: | --- | ||
Version: | 12.0-RELEASE | ||
Hardware: | amd64 | ||
OS: | Any |
Description
Viktor Dukhovni
2019-12-10 06:01:59 UTC
I have a kernel core. Don't know what to look for. Any advice on what data to extract from the savecore dump would be appreciated. Probably otherwise hard to reproduce this, as the system had been running fine for 256 days before tmpfs locked up. With the system back up I see the locked up file was a directory: # ls -ld /var/danepg/tmp/postgres/stats drwx------ 2 postgres postgres 256 Dec 10 12:35 /var/danepg/tmp/postgres/stats # ls -lR /var/danepg/tmp/postgres/stats total 72 -rw------- 1 postgres postgres 2202 Dec 10 12:35 db_0.stat -rw------- 1 postgres postgres 2033 Dec 10 12:34 db_13453.stat -rw------- 1 postgres postgres 20961 Dec 10 12:35 db_16385.stat -rw------- 1 postgres postgres 792 Dec 10 12:35 global.stat Is it likely that this is related to: https://reviews.freebsd.org/rS353065 ? Can the crash dump confirm or deny? This directory is part of a Postgres "tablespace" for temporary tables... (In reply to Viktor Dukhovni from comment #2) Sorry for the delayed reply. To start it would be useful to see backtraces from the stuck threads. If you still have the kernel and vmcore available, the output of "thread apply all bt" would be a good start. I'm sorry that this didn't get any attention when it was reported. If you still encounter this issue, "procstat -kka" output would be useful, or backtraces from a vmcore as noted in comment 3. Note that FreeBSD 12.2 is out now, please try updating if you're still on 12.0. |