Subject: Re: disk quotas - how do they work?
To: Mirko Thiesen <thiesi@network23.bsdonline.org>
From: Steven M. Bellovin <smb@research.att.com>
List: netbsd-users
Date: 08/17/2000 23:41:46
In message <Pine.NEB.4.21.0008170158260.29202-100000@ReLink.NetWork23.BSDOnLine
.ORG>, Mirko Thiesen writes:
>Hi,
>
>maybe this is a stupid questions that has been answered many times before,
>but at least *I* am unable to find the answer.
>
>I'm running a semi-public 1.4.3_ALPHA system which offers access to some
>services, including dial-in and shell access. In order to prevent my users
>from filling up my hard drives, I wanted to impose some disk
>quotas. After enabling quotas for the appropriate file systems,
>everything worked alright. But today I took a closer look at the quota
>files automatically created by quotacheck.
>
>I am just wondering if it's normal that the quota.user file on one of the 
>filesystems is rather big:
>
>(02:09:17) root@ReLink [/home] # ls -la quota.*
>-rw-r-----  1 root  operator     1048544 Aug 17 02:00 quota.group
>-rw-r-----  1 root  operator  4293918752 Aug 17 02:00 quota.user
>
>If this is unusual, may I safely remove the file and have it recreated by
>quotacheck while in multiuser mode? What else can I do?

Do 'ls -s' to find out how many blocks the file actually occupies -- 
it's almost certainly much less.  And -- if I recall correctly, though 
it's been many years since I played with the quota stuff -- repquota 
should tell you how much space each user is consuming.  I won't be at 
all surprised if there's a very high numeric uid in there.  If there 
is, you can find and remove the offending files, but that won't shrink 
your quota.user file.  But it's harmless to have it, if (as I suspect) 
most of the "space" in that file is really file system holes.

		--Steve Bellovin