Bug #1769
osd_client: susceptibility to low memory deadlocks
Status:
New
Priority:
Low
Assignee:
-
Category:
-
Target version:
-
% Done:
0%
Source:
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Crash signature (v1):
Crash signature (v2):
Description
We could be trying to flush the cache in order to free up memory, and find ourselves unable to allocate a ceph_osd or socket to (re)establish the connection through which we would flush our cache and free up some memory ... giving rise to a memory deadlock.
It would be good if OSD connection (re)establishment could be performed without additional memory allocation.
Exactly how is not clear, but Sage opines that we might pin old connections open, and then try to finish all old requests before starting new ones.
He also observes that there are probably several other such susceptibilities that should be recognized and addressed.
History
#1 Updated by Sage Weil over 12 years ago
- Project changed from Ceph to Linux kernel client
#2 Updated by Sage Weil about 12 years ago
- translation missing: en.field_position set to 27
#3 Updated by Sage Weil about 12 years ago
- translation missing: en.field_position deleted (
32) - translation missing: en.field_position set to 35