How and why does elasticsearch enter a bad state on a long query

So after running a really expensive query, elasticsearch got derped and a lot of the shards became unassigned.

They remained unassigned and I had to force assign these shards.

  1. Does this result in data loss
  2. if so, why does it result in data loss? Shouldn't elasticsearch have everything on disk for each shard. Why can't elasticsearch recover this?

What version are you on?

version 2.4

Force assigning primaries can cause data loss. Did you look into why the unassigned shards happened though - check your logs.

See for more

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.