Writes when a Couchbase Node is mark as failed

Java SDK
1

Guys,

On this Couchbase documentation page http://developer.couchbase.com/documentation/server/4.5/ha-dr/ha-dr-intro.html under the “Resilient applications” topic, it is mentioned “For example, applications can be programmed to read from replicas and to automatically retry writes on alternate nodes, with timeouts set as low as 5 milliseconds”…

I know how to handle the read case by explicitly using the readFromReplica operation…

But I am not sure how to handle the write case - I can’t seem to find a way to bypass the vBucket node owner for the write if it is down and go to another node in the SDK… Can anyone shed some light about what this sentence mean and how to achieve a resilient app when a node in the cluster is marked as failed but the failover has not happen yet, therefore rendering access to this subset of the document impossible?

Many thanks.
Regards.

2

I don’t think write is actually supported on replicas, it sounds like a documentation error…

Would you be able to raise a ticket? There’s a “Feedback on this page?” button at the bottom right to that effect. This will get routed to the relevant people by the docs team.

3

Issue reported as described.