Details
-
Improvement
-
Resolution: Fixed
-
Major
-
7.6.0
-
0
Description
Currently when we process an auto failover we block the auto_failover gen_server for the duration of the failover. Should we see a node go down whilst we process a failover (which can take some order of seconds, particularly if there are a large number of buckets to be failed over) then we will delay the failover of that node by the remaining duration of the in-progress failover. It would be beneficial for subsequent back to back failover time if we could tick auto_failover whilst we process a failover.
Attachments
| For Gerrit Dashboard: MB-59880 | ||||||
|---|---|---|---|---|---|---|
| # | Subject | Branch | Project | Status | CR | V |
| 203382,9 | MB-59880: Make failover async to auto_failover module | master | ns_server | Status: ABANDONED | 0 | 0 |
| 204763,3 | MB-59880: Make failover async to auto_failover module | trinity | ns_server | Status: MERGED | +2 | +1 |
| 205018,1 | Merge remote-tracking branch 'couchbase/trinity' | master | ns_server | Status: MERGED | +2 | +1 |
| 208979,4 | MB-59880: Short circuit failover checks if failover already in progress | trinity | ns_server | Status: MERGED | +2 | +1 |
| 209166,10 | MB-61669: Persist state on autofailover completion | trinity | ns_server | Status: MERGED | +2 | +1 |
| 209772,1 | Merge branch 'trinity' | master | ns_server | Status: MERGED | +2 | +1 |