Retention policy with mix shard type

I have a database with 60GB data, with both inmem and tsi index type, when i tried to delete from measurement, it throws error “DB contains shards using both inmem and tsi1 indexes”. I understand that I need to rebuild the index to tsi.

if I just go directly enable the retention policy without running buildtsi command, will the policy drop expired data as expected or as the same throws this error?

the concerns is how long would it take to complete the influx_inspect buildtsi conversion with 60GB and 300000 series?


How long will it take depends on the machine’s performance, but even with the data I couldn’t make an estimate.

I had a similar issue in the past, but in order to avoid rebuilding the whole index, I rebuilt single shards, buildtsi allows you to specify a single shard.
My shards were pretty small and it took only a bunch of seconds (per shard) to complete the conversion.

In order to see which shard needs to be converted you can use the report command (see here)