Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 19 Jun 2008 09:40:12 -0500
From:      Paul Schmehl <pschmehl_lists@tx.rr.com>
To:        freebsd-questions@freebsd.org
Subject:   Re: /var full
Message-ID:  <BBBB9169909ADCA813DB44EE@utd65257.utdallas.edu>
In-Reply-To: <4859FC8F.5020308@ibctech.ca>
References:  <EA09BDBE04BB7B81413DB590@Macintosh.local> <20080619035949.GB8205@shepherd>	<DB82314EA03D57CB11849EA5@Macintosh.local> <4859FC8F.5020308@ibctech.ca>

next in thread | previous in thread | raw e-mail | index | archive | help
--On Thursday, June 19, 2008 02:28:31 -0400 Steve Bertrand <steve@ibctech.ca> 
wrote:

> Paul Schmehl wrote:
>> --On June 18, 2008 11:59:49 PM -0400 Sahil Tandon <sahil@tandon.net> wrote:
>
>>> Also, what is the output of 'df -i /var'?
>>>
>>
>> # df -i /var/
>> Filesystem  1K-blocks    Used     Avail Capacity iused    ifree %iused
>> Mounted on
>> /dev/da1s1d 283737842 5397568 255641248     2%   20350 36673664    0% /var
>>
>>> See recent thread on FreeBSD Forums for context:
>>>
>>>     http://www.freebsdforums.org/forums/printthread.php?t=58071
>>
>> Thanks.  At least I know I'm not the only one to have run into this oddity.
>>
>> I'm not that knowledgeable of inodes.  My understanding is they are
>> destroyed once a file is no longer in use.  Is that correct?  Is there
>> any sort of history kept of file system activity that would identify
>> what filename was identified by the inumbers listed in dmesg.today?  Or
>> is that vain hope?
>>
>> This is a 6.2 RELEASE system.  (Looks like it's time to upgrade to 7.0
>> STABLE.)
>
> I am not in any which way certain changing major revision numbers will affect
> the file system in any which way. I am also not very knowledgeable in regards
> to inodes, but I do know that they can run out before disk space does.
>
>  From what I understand, 1MB of filespace will take up X inodes. If 1MB of
> file size is fragmented, it could take up X multiplied by N number of inodes,
> that could include a large portion of wasted whitespace.
>
> Please correct me if I am wrong.

I wouldn't be the one to correct you.  My knowledge of inodes is rudimentary at 
best.  :-)

>
> Off the top of my head, with no testing or researching behind me, what
> happens if:
>
> - stop mysqld
> - note perms of filesystem
> - cp -R /var/db /another/location/with/space
> - rm -r /var/db/*
> - fsck /dev/location-of-var
> - cp -R /copy/of/db/dir /var/db
> - reset perms
> - start mysqld
>
> ... does that free up some inodes?
>

As you can see from the df -i I posted (to which you responded), inode 
exhaustion is not an issue.  I'm leaning toward some sort of bug in mysql 
version 5.0.51 which creates a temporary file (in the wrong place) and then 
doesn't release it until it exhausts the space on the drive.  In any case, I'm 
going to report it to the mysql folks as such and hope they can figure out what 
the cause is.

-- 
Paul Schmehl
As if it wasn't already obvious,
my opinions are my own and not
those of my employer.




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?BBBB9169909ADCA813DB44EE>