Merge lp:~clint-fewbar/txzookeeper/backoff-retry-on-fail into lp:txzookeeper
Proposed by
Clint Byrum
Status: | Rejected |
---|---|
Rejected by: | Kapil Thangavelu |
Proposed branch: | lp:~clint-fewbar/txzookeeper/backoff-retry-on-fail |
Merge into: | lp:txzookeeper |
Diff against target: |
45 lines (+12/-0) 1 file modified
txzookeeper/managed.py (+12/-0) |
To merge this branch: | bzr merge lp:~clint-fewbar/txzookeeper/backoff-retry-on-fail |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Juju Engineering | Pending | ||
Review via email: mp+132113@code.launchpad.net |
Description of the change
This morning the Zookeeper server serving about 45 boxes for juju agents crashed and started recovering. They spewed ConnectionLost errors at an incredibly high rate, pounding on the zookeeper and making it worse. This code backs off the retries so the server can have some breathing room to recover.
To post a comment you must log in.
Unmerged revisions
- 52. By Clint Byrum
-
back off retry rate on connection problems to prevent dog-piling on a dead server
- 51. By Clint Byrum
-
Handle ConnectionLossE
xception more gracefully