On Fri, 2007-03-02 at 03:31 +0300, Sergey E. Koposov wrote:
> > So it seems like we already have a solution to quotas at the database
> > level.
> I still don't understand how FS quotas can help with DB quotas. ..
> All the FS quotas are setup for limiting the space for specific user and
> specific mount point. AFAIK They do not allow to limit the space in one
> directory. And since all the PG tablespaces will be owned by postgres
> user, I dont think the FS quotas can do the job.
Some filesystems help you more than others. In ZFS you can create a
filesystem that uses a pre-existing storage pool and set a quota on that
filesystem, and then mount it for the user that you want. In effect,
it's basically a directory quota (no space reservation necessary, but a
For the project you're working on, can you just set the default
tablespaces to be restricted-size areas for the scientists? If you trust
the scientists setting their default tablespace should be enough, and
they don't need to be in a separate database.
Or if you don't trust the scientists, could you use a separate database
and the scientists could query with something like dblink?
In response to
pgsql-hackers by date
|Next:||From: Jim C. Nasby||Date: 2007-03-02 03:30:49|
|Subject: Re: Revitalising VACUUM FULL for 8.3|
|Previous:||From: Sergey E. Koposov||Date: 2007-03-02 01:22:09|
|Subject: Re: SOC & user quotas |