jackrabbit-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gary Webster <webs...@lexmark.com>
Subject Re: [Jackrabbit] runaway Postgres bundle Toast table
Date Wed, 01 Aug 2012 21:50:55 GMT
Hello.

I've now noticed that Jackrabbit is doing this "update ... BUNDLE_DATA = $1
..." for each file change/add, where $1 is 6MB+ in size !
Surely that is not functioning as designed.
I could easily saturate a 100Mb ethernet with just this statement!

Can anyone give me some advice on this?
Thanks.

On Mon, Jul 30, 2012 at 12:41 PM, Gary Webster <webster@lexmark.com> wrote:

> Hello.
>
> I am currently experiencing a horrible issue, using Jackrabbit clustering
> with PostgreSQL as the Persistence Manager.
>
> This particular table (ws_bundle) grows ~10GB/hr !  My entire DB, for
> multiple apps, including the rest of the repository DB, is <1GB.
> This table is even bigger than the data store!
>
> I enabled query/statement logging in Postgres.  The log went to 1GB in
> just a couple minutes!
> It is dominated by this one statement, which occurs every ~2 sec:
> "update WS_BUNDLE set BUNDLE_DATA = $1 where NODE_ID_HI = $2 and
> NODE_ID_LO = $3"
> Parameter $1 is hex & appears to keep growing exponentially.  The last one
> I could catch was over 7million characters long !!
> I can't believe that is correct functionality.
>
> Surely, I have something wrong in my Jackrabbit-db configuration.
>
> Thanks for your help.
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message