Backport #18132
hammer: ReplicatedBackend::build_push_op: add a second config to limit omap entries/chunk independently of object data
Release:
hammer
Crash signature (v1):
Crash signature (v2):
Related issues
History
#1 Updated by Alexey Sheplyakov over 7 years ago
- Copied from Feature #16128: ReplicatedBackend::build_push_op: add a second config to limit omap entries/chunk independently of object data added
#2 Updated by Nathan Cutler over 7 years ago
- Status changed from Pending Backport to New
#3 Updated by Nathan Cutler over 7 years ago
- Description updated (diff)
original description¶
In ceph deployments with large numbers of objects (typically generated by use of radosgw for object storage)
it's quite possible for OSDs recovering data to hit their suicide timeout and shutdown because of the number
of objects each was trying to recover in a single chunk between heartbeats.
https://bugs.launchpad.net/ubuntu/+source/ceph/+bug/1628750/comments/0
build_push_op assumes 8MB of omap entries is about as much work to read as 8MB of object data. This is probably false.
Add a config (osd_recovery_max_omap_entries_per_chunk ?) with a sane default (50k?) and change build_push_op to use it.
#4 Updated by Nathan Cutler over 7 years ago
- Description updated (diff)
- Status changed from New to In Progress
- Assignee changed from Alexey Sheplyakov to Nathan Cutler
#5 Updated by Nathan Cutler about 7 years ago
- Status changed from In Progress to Resolved
- Target version set to v0.94.10