Description
The alert "Clock CAS drift threshold exceeded for bucket" is potentially misleading to an end user. The relationship between CAS (used for concurrency operations) and XDCR is non-obvious to an end user. The message would be improved by simply dropping "CAS"
Further suggestions to improve:
- Modify the text to include the possible impact and how to resolve (or a permalink)
- Include which node name(s) the discrepancy was detected on.
- Include what the threshold is
e.g. A time difference greater than 5 seconds was detected on node(s) 123.123.123.123. This can affect time-based conflict resolution. Clocks should be synchronised as per <some link>
Attachments
Issue Links
- relates to
-
MB-21153 Alert when |max_cas - wall_clock| > exceeds threshold
- Closed
For Gerrit Dashboard: MB-21538 | ||||||
---|---|---|---|---|---|---|
# | Subject | Branch | Project | Status | CR | V |
69936,2 | MB-21538: support more flexible stat-based alert messages | watson | ns_server | Status: MERGED | +2 | +1 |
69937,3 | MB-21538: improve drift ahead threshold exceeded alert message | watson | ns_server | Status: MERGED | +2 | +1 |
69984,1 | Merge remote-tracking branch 'couchbase/watson' | master | ns_server | Status: MERGED | +2 | +1 |