ILM Policy poll interval

Hi all,

In our project environment, we have changed the Cluster ILM poll interval from default (10m) to "1s". Does we face anyindices performance related issues on clusters ?

Note : Our cluster size is 3 node. Also need to achieve indices rollover without lag in the max index size, Max Docs Count and Time frame .

PUT /_cluster/settings
"persistent": {"indices.lifecycle.poll_interval": "1s"
} '

Could anyone suggest the better ILM poll interval in production env cluster ?

I would not recommend 1 second for production, simply because as the number of indices grows, they'll need to be checked every second, which is usually more load than is necessary (1 second is usually used when you're testing something out, or during development).

As far as a better default, I'm surprised that 10m isn't reasonable for you. Are you ingesting such a large amount of data that it's exceeding the size you want to see within ten minutes? Can you explain more about why that isn't working for you?

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