Elasticsearch index corruption?

[Ramon_Zaro (Ramon)]

Hi, I am running a ES 6.3 on a single node fed by fscrawler 2.4 on a fileshare. It was working fine for few days but suddenly there was an issue that certain keywords would give a timeout (searched using kibana 6.3) or a no response (.NET client using ES6.3 rest service).

I am really perplexed as to the cause, the node has 40 shards of which 5 are unallocated, thus health is showing yellow.

My questions are :
a) has the index been corrupted? is there a way to check that ?
b) if so, can it be recovered ?
c) what are the reasons that can lead to such corruption of index and how to avoid them?

thanks in advance.

[warkolm (Mark Walkom)]

It'd be best if you have a look in your Elasticsearch logs before jumping at corruption concerns.
Can you find them and post a few excerpts?

[Ramon_Zaro (Ramon)]

Hi @warkolm, I did have a look at the logs and they frankly didn't have anything of note.

Other than some g.c messages there was absolutely nothing. slowlogs were empty.

Please let me know if there's something else I can have a look at to gauge what's going on.

Currently, some keywords return a result while others timeout; f.e 'vanderbilt' times out, while 'van' & 'van*' returns results.

[Ramon_Zaro (Ramon)]

@warkolm, anyone ? anything ?

I saw a few issues reported with G.C, does that have anything to do with this ?

[warkolm (Mark Walkom)]

Perhaps if you show your query.