Actions
Bug #55734
openrgwlc: try harder to avoid lock starvation when storing new lifecycle policy
% Done:
0%
Source:
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Description
The current logic in guard_lc_modify(...) attempts to take an lc shard lock in a non-terminating do-while loop, which is not ideal. Also, it sleeps 5s between attempts. Address both issues, so that the call tries harder to outcompete other call paths, but also times out after a reasonable period (e.g., 50s, about the length of time a typical client would wait).
Updated by Casey Bodley almost 2 years ago
- Status changed from In Progress to Fix Under Review
- Pull request ID set to 46248
Actions