Actually, that solution didn't work so well. Even very small delays in the loop caused the entire loop to perform too slowly to be useful in the production environment. I ended up producing a small patch out of it :P, but we ended up using pgpool to reduce connections from another part of the app, which made the pg_autovacuum spikes less troublesome overall.


Thomas F. O'Connell
Co-Founder, Information Architect
Sitening, LLC

Strategic Open Source: Open Your i™
110 30th Avenue North, Suite 6
Nashville, TN 37203-6320

On May 15, 2005, at 9:26 PM, Matthew T. O'Connor wrote:

Mindaugas Riauba wrote:

The "vacuum cost" parameters can be adjusted to make vacuums fired
by pg_autovacuum less of a burden. I haven't got any specific numbers
to suggest, but perhaps someone else does.

It looks like that not only vacuum causes our problems. vacuum_cost
seems to lower vacuum impact but we are still noticing slow queries "storm".
We are logging queries that takes >2000ms to process.
And there is quiet periods and then suddenly 30+ slow queries appears in
log within the same second. What else could cause such behaviour? WAL log
switch? One WAL file seems to last <1 minute.

How long are these quite periods? Do the "strom" periods correspond to pg_autovacuum loops? I have heard from one person who had LOTS of databases and tables that caused the pg_autovacuum to create a noticable load just updateing all its stats. The solution in that case was to add a small delay insidet the inner pg_autovacuum loop.

---------------------------(end of broadcast)--------------------------- TIP 4: Don't 'kill -9' the postmaster

Reply via email to