On Mon, May 13, 2013 at 8:37 AM, S H <msq001@live.com> wrote:
>> What is the db server doing when this happens? What does top, vmstat,
>> iostat etc have to say about it?
>
> It is high end server with 96 GB of RAM , 16 core server, but there are many
> other application running, This db is used for monitoring the performance of
> this server and inserting/updating data every one in 10-20 tables.
>
> I am able to reproduce this issue, in case i run vacuuming of my database
> and in parallel try to open 30 connections, sometime db client takes time to
> open connection .... Is there any workaround or there is some known issue
> already existing. If there is any known issue it will be easy to persuade my
> customers to upgrade..
I wonder if you've got a bloated pg catalog then. Certainly sounds
like it's a possibility.
So other than vacuuming when you recreate this, is the server working
hard? What is vacuum vacuuming when this happens (pg_stat_activity
should show that).
>> Running 8.1 means you're asking about a system no one else on this
>> list is likely to still be using much. The hackers aren't gonna be
>> interested in fixing it either, since it's out of support.
>
> I am migrating to new version, for new customers , but for old existing
> customer, it would require significant time. Is there any workaround for the
> same. Like improving some DB parameters.
Probably not. If the tables that make up the pg catalogs are bloated,
you might have to go take some outage time to run a vacuum full /
reindex on them. Some of them, I believe, require single user mode to
do this, but not sure, and definitely not sure if it's na 8.1 thing or
not.
