Version : ES 7.6.0 I have elastic search set up with filebeat for pulling up logs and rollover configured with index policy(ILM) . I have 2 phases configured : HOT and DELETE. In the delete phase for index policy there is option in kibana to delete based on number of days from rollover which i am having as 3 days. I recently faced an issue wherein load test was performed and there was sudden spike in disk usage,index policy did not delete the indices as rollover configured was 3 days . As a result there were watermark alerts after 95 % and index went into read only mode. My question is for such a scenario is there a way in index policy to delete old indices after a certain threshold of disk space is utilized instead of just configuring deletion based on number of days from last rollover of index?
There's no disk level based trigger at this stage, check out https://github.com/elastic/elasticsearch/issues/49392 though.
Thanks for your reply. I am surprised that this feature is not implemented yet as this seems to me a fairly common use case . So how to deal with such scenarios wherein there could be space crunch and I wish to delete old indices to accomodate space for new ones?
You would currently need to monitor it yourself.
Curator does support this as far as I know.
On the same lines , is there a feature where i can bind ES to use certain amount of space only on a disk like for eg 50 GB to ES on 200 GB disk ?
So value in "cluster.routing.allocation.disk.watermark.flood_stage"
will be the max value ES will use on the node .
Is this understanding correct ?
Yes that's right.
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.