Thread: vacuum job taking very long time to complete
Hi,
I've started a vacuumdb on a database having 2 large tables of approx. 3,800,000 records each. The database size is approx . 2Gbyte
It generates a lot of logfiles, currently the pg_xlog directory has about 4.6 Gbyte of logfiles and when I started there were I guess 2 logfiles each 16 Megs.
I still have 3.5 G available for logfiles but suppose there is a chance that I will run out of diskspace before the vacuum is done , can I kill it safely before that happens ?
Thanks,
Feite Brekeveld <feite.brekeveld@osiris-it.nl> writes: > I still have 3.5 G available for logfiles but suppose there is a chance > that I will run out of diskspace before the vacuum is done , can I kill > it safely before that happens ? Yes, you can just send a SIGINT to the vacuuming backend (or type control-C in psql, if you issued the vacuum command from psql). You may care to apply the patch in http://www.ca.postgresql.org/mhonarc/pgsql-patches/2001-06/msg00061.html before trying again. regards, tom lane
Tom Lane wrote: > Feite Brekeveld <feite.brekeveld@osiris-it.nl> writes: > > I still have 3.5 G available for logfiles but suppose there is a chance > > that I will run out of diskspace before the vacuum is done , can I kill > > it safely before that happens ? > > Yes, you can just send a SIGINT to the vacuuming backend (or type > control-C in psql, if you issued the vacuum command from psql). What happens with the logs when you do that ? Are the cleaned up because of SIGINT ? > > > You may care to apply the patch in > http://www.ca.postgresql.org/mhonarc/pgsql-patches/2001-06/msg00061.html > before trying again. > > regards, tom lane -- Feite Brekeveld feite.brekeveld@osiris-it.nl http://www.osiris-it.nl
On Wed, Jun 27, 2001 at 04:10:05PM +0200, Feite Brekeveld wrote: : Hi, : : I've started a vacuumdb on a database having 2 large tables of approx. : 3,800,000 records each. The database size is approx . 2Gbyte I'm seeing a similar slowdown, but on much smaller tables (roughly 10,000 and 30,000 rows apiece). It takes up to a minute to analyze the 10,000 row table (see other messages to this list about the constantly updating nature of the application using this database). The weird thing is that these vacuums took barely any time before a recent required restart of the database. Nothing's changed in the config, yet now the vacuums take a while now. I'm not sure what to make of it. The system doesn't appear to be under any sort of increased burden (in fact, postgres is using barely any resources during the analyze). As Austin Powers would say, "That's not right ..." * Philip Molter * DataFoundry.net * http://www.datafoundry.net/ * philip@datafoundry.net