Bug #10241
Incorrect OSD mapping with EC 6+2 setup in Giant
Status:
Resolved
Priority:
Normal
Assignee:
-
Category:
-
Target version:
-
% Done:
0%
Source:
other
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):
Description
Hit this on the performance test cluster during nightly giant testing. Notice the very incorrect mapping.
perf@magna031:/tmp/cbt/ceph/log$ ceph --version ceph version 0.87-44-g26e8cf1 (26e8cf174b8e76b4282ce9d9c1af6ff12f5565a9)
perf@magna031:/tmp/cbt/ceph/log$ ceph health detail HEALTH_WARN 1 pgs degraded; 1 pgs stuck degraded; 1 pgs stuck unclean; 1 pgs stuck undersized; 1 pgs undersized pg 12.244 is stuck unclean since forever, current state active+undersized+degraded, last acting [8,6,2,0,10,2147483647,1,11] pg 12.244 is stuck undersized for 23610.100317, current state active+undersized+degraded, last acting [8,6,2,0,10,2147483647,1,11] pg 12.244 is stuck degraded for 23610.100408, current state active+undersized+degraded, last acting [8,6,2,0,10,2147483647,1,11] pg 12.244 is active+undersized+degraded, acting [8,6,2,0,10,2147483647,1,11]
History
#1 Updated by Sage Weil over 9 years ago
- Status changed from New to Need More Info
need osdmap or crushmap that triggers the failed mapping
#2 Updated by Loïc Dachary almost 9 years ago
- Status changed from Need More Info to Resolved
- Regression set to No
It needs 8 and there are only 9 available, which may be a problem for crush, sometimes. It has been resolved (or made very rare) by increasing the number of tries from 50 to 100 for erasure code crush rulesets a while ago.