Actions
Bug #17174
closedWatcher not removed on host reboot
Status:
Rejected
Priority:
High
Assignee:
-
Category:
librbd
Target version:
-
% Done:
0%
Source:
other
Tags:
Backport:
Regression:
No
Severity:
2 - major
Reviewed:
Description
Hello,
I'm currently running Jewel 10.2.2 (45107e21c568dd033c2f0a3107dec8f0b0e58374) and ran into an issue where on a host failure/power off the images that are currently mapped on that host do not lose their watchers.
Steps to reproduce:
- Map and mount volume
- Kill node
- Wait...
It does appear that after 12+ minutes that the watcher ended up being removed, but my assumption is that the watcher should be removed after the 30s timeout for client crashes
Regards,
Connor
Updated by Samuel Just over 7 years ago
- Status changed from New to Need More Info
There are cases where it can take longer. Can you reproduce with
debug osd = 20
debug filestore = 20
debug ms = 1
on the primary for the osd responsible for one of the objects?
Updated by Sage Weil almost 7 years ago
- Status changed from Need More Info to Rejected
Actions