launchpad-dev team mailing list archive

[William Grant <william.grant@xxxxxxxxxxxxx>]

On 25/01/12 09:27, Deryck Hodge wrote:
> Hi, Abel.
> 
> We've begun to run out of disk space on devpad due to all the oops
> piling up.  William has kindly offerred to try to get a fix for this
> in his morning today, in order to get us some disk space back.  If he
> can't fix it, he'll do some manual oops deleting, but obviously,
> fixing it is better than blindly deleting old oops.
> 
> I just wanted to give you (and the list) a heads up, so others know
> what's going on with the issue and so you know to check on the state
> of the fix before diving back in during your morning tomorrow.

As I've described in the bug, it's a case-sensitivity issue. Launchpad
returns referenced OOPS IDs in ALLCAPS, while oops-datedir-repo treats
them case-sensitively. I have a branch in ec2 to fix Launchpad.

But since the fix cannot wait for a deployment in a day or two, I've
cowboyed oops-datedir-repo on carob to treat OOPS IDs insensitively.
It's currently pruning over the production OOPSes, having so far removed
16GB of about 87GB. I've kept a hardlinked snapshot of the original tree
in lpqateam@carob:~/oops-amqp-backup just in case, so the space won't be
reclaimed until I've verified that the pruning was correct.

Attachment: signature.asc
Description: OpenPGP digital signature