Exception logging, shard failures multisearch

Hi All,

Main problem:
Recently we started testing the #succesful shards==#total shards while
doing a multisearch.
After some time the application alarms that a multisearch had failing
shards.

It seems to be time related, since, if we repeat the query, it succeeds.
No errors in the logs, just 1 failed shard was mentioned in the response.
The cluster is green at the time of query.

Exception logging
I noticed that I have to set the log level to TRACE to get exception
logging.
To me this seems a bit weird, because exception logging is very important,
and TRACE has the same level as DEBUG...
So, now I not only get my errors logged, but also I also get lots of
messages that I didn't ask for :frowning:

After putting the log level to TRACE, we still don't see errors in the log.
There is only the #failed shards==1 in the response...

Any ideas?

Kind regards,
Peter

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

Update: I can reproduce the behavior with a multisearch with only 1 query.
(with a lot of requesting clients)

I know the #hits that the query returns when it doesn't fail: 44
When the multisearch reports 1 failing shard, the #total hits is correct
number. (44)
When I set the size of the query to be >= the #total hits, I see the
results from the last shard missing.

Also: there is no "failures": element in the ressponse.

I forgot to mention: I'm using ES 0.90.1

Peter

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.