Eternal DataFileRead on partition seeks after upgrade? Or in general..
I have a table that is partitioned by month. It contains rows of plays per a game, with about 800k rows per game, and maybe about 350 games per month. Its PK is a game identifier, a play identifier, and the game's date as that is partition key.
It's a lot of data, I think.
Before upgrading to 13, it used to be a reasonable amount of time to pull records based on multiple game identifiers (even if it was a few million rows).
Since upgrading to 13, it seems it is endlessly waiting on DataFileRead to query this table, and things take hours.
I'm wondering what I can do: is the bottleneck truly the disk I/O (it wasn't really a disk issue before).
I did run vacuumdb --analyze-only after upgrade, but if doing anything else: re-indexing the PK, etc, might help these queries seek the proper partition more quickly (guessing that is what is happening with the DataFileRead), I'd be eager to learn.