We have this setting in our elasticsearch.yml configuration file:
indices.lifecycle.poll_interval: 30s
And we have a set of indexes that must roll over every 3 minutes and must be alive for 6 minutes.
The rollover is happennig on time, but the deletion only occurs exactly every 10min.
For example:
01:52:30: index idx-000001 created
01:55:55: index idx-000002 created
01:59:25: index idx-000003 created
02:02:30: index idx-000001 deleted
So it seems the 'indices.lifecycle.poll_interval' setting correctly informs about rollover actions every 30s, but not about delete actions (that is taking 10min, the setting default value).
Is this possible? Must I set up another setting for delete actions? Or am I doing something wrong?
Can you share your actual policy? It's a little harder to parse from this sentence, but I suspect that you have a min_age of 6 minutes in your delete phase, which means that it will be deleted 6 minutes after the rollover has been performed (so 3 + 6 = ~9 minutes total lifetime)
Yes, you are right. That's the point, deletion takes rollover as initial point and not the index creation.
I need to delete indexes taking the index creation as a base. So I have decided to delete them manually. I understand that deleting indexes manually does not interfiere in the ILM.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.