node timeout on overwrite can easily cause mis-matched etag fragment to 503
Bug #1484564 reported by
clayg
This bug report is a duplicate of:
Bug #1457691: node timeout on overwrite can easily cause mis-matched etag fragment to 503.
Edit
Remove
This bug affects 1 person
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| OpenStack Object Storage (swift) |
Incomplete
|
Low
|
Unassigned | ||
Bug Description
Anytime a object PUT overwrites and object, and one of it's fragments is placed on a handoff, the next immediate read (if the primary node that was skipped is back online, and iff the read is before reconstructor can fix it) will 503 because one of the successful responses from the backend node will have a fragment from the old object.
This happens more than you would like when doing benchmarking.
Luckily there is fix => https:/
Revision history for this message
| paul luse (paul-e-luse) wrote | #1 |
Revision history for this message
| clayg (clay-gerrard) wrote | #2 |
| Changed in swift: | |
| status: | Confirmed → Incomplete |
| importance: | Critical → Low |
To post a comment you must log in.
FYI was tracking the fix to https:/