Hi Klaus,
Am 08.05.2015 um 18:38 schrieb Klaus Thorn <[email protected]>:
> Dear archiveopteryx users,
>
> currently we see a problem at our Archiveopteryx 3.2.0 installation:
>
> more and more postgres queries from Aox queue up and sooner or later
> cause resources to be too scarce for other services.
>
> The queries are all alike, here one example with a bit context from the
> log file:
>
> debug: 2/5801/3/172/6: 2015-05-08 10:53:01.226 Sent parse/execute for
> "select mm.uid, mm.modseq, mm.message
> from mailbox_messages mm
> where mm.mailbox=$1 and mm.uid=any($2) and not mm.deleted"
> ($1='441',$2='{16480,16481'... (2.9Mb)) on backend 32
>
Seems to be a backend tuning issue.
Try explain analyze verbose select ā¦ in psql and add spindles/RAM as
appropriate. (-;
> Our biggest mailbox has 441407 messages.
Let me guess: The problem happens if queries about one of these big mailboxes
are queued up.
Iām not aware of any MUA which can handle such amount of messages in a single
mailbox.
Axel
---
PGP-Key:29E99DD6 ā +49 151 2300 9283 ā computing @ chaos claudius
