benny@bennyvision.com ("C. Bensend") writes:
> Hey folks,
>
> I'm running 8.0.4 on OpenBSD, and I'm running into issues where
> a large batch job will fire up and occasionally coincide with a
> VACUUM. Which then makes this batch take an hour and a half,
> rather than the 30 minutes it usually takes. :(
>
> I am going to upgrade to 8.1.2 this weekend, before you ask. :)
>
> Here's the question - is there a query I can make in my scripts
> (using perl/DBD::Pg) to see if the database is being VACUUMed at
> the current time? I could add a sleep, so the script would
> patiently wait for the VACUUM to finish before kicking off several
> intense queries.
>
> Thanks much!
If you have command string monitoring turned on, via
stats_command_string in the postgresql.conf file, then you could get
this information from the system view pg_stat_activity.
Generally, you could look to see if a current_query is a vacuum, perhaps via...
select * from pg_stat_activity where lower(current_query) like 'vacuum%' ;
If that parameter is not turned on, then ps auxww | egrep [something
finding your PG processes] | grep VACUUM could perhaps do the trick,
albeit not from a straightforward database query...
--
output = ("cbbrowne" "@" "acm.org")
http://www.ntlug.org/~cbbrowne/advocacy.html
"Is your pencil Y2K certified? Do you know the possible effects if it
isn't?"
