During my CB-dev (relatively small), i’ve caught a couple (ok, maybe 3-4-5 ) bugs that was, imho, related to “instability of secondary indexes”. Two times those were confirmed bugs ( [moved: JavaSDK->CouchbaseServer] N1QL for 4.1.X count(*) problem , Index problem with flush + select ... [request_plus | not_bounded]), but there is something, that i can describe like “sporadic failures with secondary indexes after index-change-related processes”. Flush and node rebalance are the ones (further investigation needed, at least for rebalance; for flush it seems like https://issues.couchbase.com/browse/MB-16957; i hope, this is the only one). But there is something more, like “secondary indexes sporadically just fails with simple SELECT” (even with scan_consistency=NOT_BOUNDED), at least for 4.1.0 .
Has anyone got something like this?
Do you have the error you are getting from the SELECT that is returning errors.
there is nothing “confirmed” for now. I need to investigate further, but i just feel myself little bit drained (at least 3 discovered bugs for last couple of weeks, all were “investigated and confirmed” and at least 2-3 [as i suspect] more was not even published on this forum). So this topic is addressed to developers who possibly caught something like that. Their answers, probably, could make my investigations easier.
ok - happy to take a look at the logs too. we appreciate your contributions. Pls keep it coming.