>>>>> "BM" == Bruce Momjian <pgman@candle.pha.pa.us> writes:
>>
>> Well, did I mention I'm saturating my disk I/O bandwidth at the same
>> time with other queries? ;-)
BM> But six hours. It is my understanding that a sequential scan is roughly
BM> the same load as a non-FULL vacuum. Are you saying a sequential scan
BM> takes +6 hours too? How can any work get done?
Well, it is a big downward spiral once you saturate your disks. You
can't get queries done quickly enough, and table size increases with
all those updates, and you have to run vacuum to help that, and that
slows the queries more. Lather rinse repeat.
The new server, with 14 ultra-160 SCSI disks on a hardware RAID10 is
going to go to the data center tomorrow, and hopefully will be live by
the weekend (if I can get eRServer to replicate the data to the slave
without having to shutdown the whole system).
Then I'll let ya know how long the vacuum takes ;-)
PS: Last time I tried a vauum full on my largest table, I gave up
after 14 hours of down time. That was not good for our business...