Hi there
we're going through our own ILM components and noticed there are a number of built-in ILMs. One of these is for the watcher history indexes such as ".watcher-history-12-2020.12.23".
I'm thinking these are mis-configured? or there is some type of configuration I'm not understanding.
we are running 7.10.1
Here is what I see:
- ILM Policy: watch-history-ilm-policy
- this policy is connected to zero (0) index templates
- I had edited the default ILM policy to execute a Delete after the index was 30 days old.
- There are no index templates (managed or legacy) for *watch* indexes
- The watch index itself contains the following setting
- "settings": { "index": { "lifecycle": { "name": "watch-history-ilm-policy"
- But the rollover_alias setting for that index is blank ("")
I'm not sure what connects this index to the lifecycle management policy other than the name above - but this is what were getting in our logs:
[elasticsearch.server][INFO] policy [watch-history-ilm-policy] for index [.watcher-history-12-2020.12.24] on an error step due to a transient error, moving back to the failed step [check-rollover-ready] for execution. retry attempt [59]
[elasticsearch.server][ERROR] policy [watch-history-ilm-policy] for index [.watcher-history-12-2020.12.24] failed on step [{"phase":"hot","action":"rollover","name":"check-rollover-ready"}]. Moving to ERROR step
Perhaps i'm not understanding how ILM works with out-of-the-box indexes? Or perhaps the ILM policy connection to the index isn't really there but since there is a policy name linked then ILM execution components think it should do 'something'.
I'm also wondering if they failure is because of the additional delete i put into the policy?
Thoughts?
thanks,
Michael