Project

General

Profile

Bug #58013

Osdmap too big lead osd crash

Added by 伟杰 谭 3 months ago.

Status:
New
Priority:
Normal
Assignee:
-
Category:
-
Target version:
-
% Done:

0%

Source:
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
rados
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

My cluster has failed, a large number of osds cannot be started,
and the troubleshooting found that the size of osdmap is too large, causing osd to start abnormally

[root@osd-36 ~]$ ls -l osdmap_503314
-rwxr-xr-x 1 root root 39170644 Feb 24 2022 osdmap_503314

osdmap looks normal, and there seems to be nothing wrong with executing the command output
[root@Rosd-36 ~]$ osdmaptool osdmap_503314 --print|more
osdmaptool: osdmap file 'osdmap_503314'
epoch 503314
fsid 46b2bef0-9d6c-49eb-a147-083fb0cd18d8
created 2020-04-15 17:45:12.738276
modified 2022-11-12 17:36:31.661536
flags noout,noscrub,nodeep-scrub,sortbitwise,recovery_deletes,purged_snapdirs
crush_version 901
full_ratio 0.95
backfillfull_ratio 0.9
nearfull_ratio 0.85
require_min_compat_client jewel
min_compat_client jewel
require_osd_release mimic
....

So is it normal to say that the map reaches 38M?

Also, how to upload this osdmap sample if needed

Also available in: Atom PDF