Hello,
Replication was enabled between 2 bucket via xdcr. Once the operation completed, I notice the item count does not match.
source:138,060,006
destination:138,060,004
This is not the first time this happened, first time I was missing 2 million rows. But there was a lot of activity on both clusters. I imagine it was asking too much. So I retested the process again, this time both cluster basically sleeping.
Questions:
Does someone wrote an application to sync missing rows between 2 buckets? If so, do you have info on the program like a GitHub link?
Is there a particular line I could monitor in couchbase logs to detect when I have an issue replicating data between two buckets?
-Steeve
You don’t indicate the version, but if it’s a recent one, have a look at the goxdcr log. If it’s 4.0 or earlier, have a look at xdcr_errors or xdcr. The documentation covers this and indicates log locations.
I know about the logs, that is why I asked about what line in the logs. This is a 5 nodes cluster. With 7 hours of logs. I wanted to know if I can look for a specific line in the log that would indicate why my document was not properly synced.
Are you able to identify which two documents are missing between the two clusters?
Also, have you performed any deletes on the remote bucket at all?
Deleting a document will increase its revision ID by 1 and due to the way XDCR conflict resolution works it will lead to the document not being replicated from the source bucket. It is quite common for this to be the cause in the case of a few documents difference between the buckets.
Additionally, failure to replicate to the remote bucket should not cause items to not be processed/replicated as all failed mutations are re-queued internally.