Problem
We have set the following index management policy for indices, which moves the index to delete phase after 32 days & then once the snapshot is completed the index will be deleted.
{
"policy": {
"phases": {
"hot": {
"min_age": "0ms",
"actions": {}
},
"delete": {
"min_age": "32d",
"actions": {
"delete": {
},
"wait_for_snapshot": {
"policy": "recurring-snapshots"
}
}
}
}
}
}
Recently, all the indices got deleted before they reached the min_age
i.e 32 days. However after restoring the data from the snapshot the ILM is again working fine. We checked the trigger for index deletion & it was indeed done by the same policy and wasn't a manual trigger. Checked this through logs.
Debugging approach
- Check when the index was deleted through logs
- Check through
.ds-ilm-history-*
index pattern for the trigger of ILM moving the index to delete phase, also checked the phase_definition on the document given here as an output the definition wasn't modified.
Questions.
- Is there a way to find how did such a event got triggered?