CBG-4416 Use waitForTombstoneVersion to validate tombstone versions #7252
Conversation
Validates CV for deletes.
| if actualTargetCas != 0 { | ||
| base.DebugfCtx(ctx, base.KeyVV, "Target document exists as a tombstone, but does not have _vv, _mou, _sync") | ||
| } |
There was a problem hiding this comment.
Is this useful to leave in debugging? At this point it might or might not get replicated depending on conflict resolution, despite having no _vv
There was a problem hiding this comment.
I left it in as it's a bit of an unusual case that I don't expect us to hit in future (when we've removed the need for the dummy xattr), and it would be useful to see when it continues to show up in that case.
topologytest/hlv_test.go
Outdated
| t.Logf("Don't include deleteVersion from Couchbase Lite peers when determining lastWrite %s, CBG-4432", peerName) | ||
| continue | ||
| } | ||
| t.Logf("deleteVersion on peer %s: %+v, implicit version: %+v", peerName, deleteVersion, deleteVersion.ImplicitHLV) |
There was a problem hiding this comment.
Is implicit version helpful to log here? I think logging with %#v will log all the details. Whether or not something has an implicit hlv only indicates whether it was written by couchbase server, not via XDCR.
There was a problem hiding this comment.
Agree - will remove the redundant implicitVersion..
Validates CV for deletes. Uptakes rosmar fix and includes a few test tweaks to ensure CBL version isn't used for lastWrite until CBL deletes are implemented in test framework.
CBG-4416