On 26 May 2011, at 15:19, Daniela Bauer wrote:
> Hi All,
>
> In the past week, I've twice managed to kill my WMS/LB setup by filing
> up the space allocated to the mysql databases.
> The database on the WMS is currently ~8GB and growing and I haven't
> found a way yet to clean it - does anybody know how to do this, short
> of a reinstall ?
> Also on the (glite 3.2) LB the automated purge seems to fail (I
> submitted a ticket about this in February) and cleaning it by hand
> takes ages - the database now has reached 20GB and the way it is going
> the system is going to last two more weeks before croaking again.
>
> Any hints gratefully received.
We've taken the easy way out, and allocated acres of space to mysql (i.e. the 'disk is cheap' philosophy). At the moment, we've got 58GB in one, and 69GB in the other [0]. It's actually not a terrible option for the LB's, as they are well indexed (unlike some other services), so the bulk of the database does nothing more than take up disk space.
I've poked through the LB database before - from memory most of the jobs are marked finished and can be removed; but I recall there being a script to do that kicking around already. In the event that one's not forthcoming, and I can dust off what I know, and come up with something that should help.
[0] Although part of that is binary logging, as per: http://scotgrid.blogspot.com/2009/08/database-backups-and-lock-time.html so it's 42GB and 55GB actual database.
|