Hi. We have archived and removed majority of data from a database, the
main impact was on 4 tables, which lost several million rows (3
tables) and several dozen million rows (one table).
Naturally we decided to execute VACUUM FULL on the database to reclaim
all the space; it keeps running for 22 hours already.
Can VACUUM get stuck and run forever e.g. should we cancel it?
Is there any way to diagnose the amount of time remaining for it to
run?
Should we rather cancel it, and run a normal VACUUM? 8.2 manual said
you should use FULL when the majority of data is deleted but I'm
concerned about the time it takes :)
Also, would increasing maintenance_work_mem while the query is running
help? It's currently 2Gb out of 4Gb RAM, postgres proccess supposedly
running the vacuum is slowly eating up memory but it's not even at 30%
yet...