[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Filled up the filesystem. How?

Dotan Cohen wrote:
On 6/8/05, Alexander Dalloz <ad+lists uni-x org> wrote:

Am Di, den 07.06.2005 schrieb Dotan Cohen um 23:39:

Last week I wrote that I somehow filled 7 out of 10 megs in my linux
partition. Today that last bit was filled- I am at 100% capacity.

I cannot download email or create new files. What could be the cause
of this? Where should I look for bloat? What can I delete?


This can easily be happen if log files fill very quickly. I.e. if you have Apache running, a fault in your page and quite some hits, the error_log can grow rapidly. So watch out for large log files.


-- Alexander Dalloz | Enger, Germany | GPG http://pgp.mit.edu 0xB366A773 legal statement: http://www.uni-x.org/legal.html Fedora Core 2 GNU/Linux on Athlon with kernel 2.6.11-1.27_FC2smp Serendipity 23:54:54 up 14 days, 22:32, load average: 0.38, 0.53, 0.49

BodyID:69189987.2.n.logpart (stored separately)

/var/logs is 23 megs (same as last week) /var is 1.3 gigs (same as last week)
/usr is 3.7 gigs (same as last week)
/proc is 480 megs (same as last week)

I only checked those because those were the biggies last week. The
system is so slow now that it takes a long time for it to caculate
those values. Where else should I look?


Dotan -

Can you show us an output of:

/sbin/fdisk -l|df -i

The reason I ask is ebcause I've seen a machine that had been out of inodes, as seen by this output. Pay attention to the "IUse%" column, which may provide clues. If this is the case, /var/log/messages will hint at this.

When a disk is formatted, each unit is designated an allocation size. There is one inode for this allocation block size. If your block size is, for example, 4096k, and you have a file that is 2k, then you will have used one inode. You can see how this would be a problem if you had hundreds of thousands of very small files, which would make your system *think* that it's low on space, just because inodes are not being used efficiently. I've seen this many times on compromised mail servers where a spammer has queued up millions of tiny email messages.

Hope that helps

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]