Just found out that one of the existing data sources couldn’t be query and that actually /druid/indexCache/$data_source_name dir doesn’t exists, while segments are still active in /druid/localStorage/ and in the database.
Anyone has a clue on what could be the cause if this?
Note that particular data source haven’t been ingested new data for about two months.
Maybe there’s a drop rule for that datasource that drops data after two months?
Thanks for the reply.
That was one of my doubts too, but wondering whether maybe Druid imposing such scenario by default (or something).
However, how to re-create cache files in /druid/indexCache/$data_source_name and make data source queryble again (there are still segments and meta data available)?
Druid doesn’t drop data by default, but you might have configured a rule and forgot about it.
Are the segments still marked “used” in the metadata store (used=true)? If so then historicals should be automatically loading them. If they’re not, there’s probably something wrong with the loading process.
If the segments are not marked “used=true” then that means someone, or a drop rule, dropped them. You can get them back by disabling the drop rule if it exists, and re-enabling the datasource using the enable datasource api.