file system full

Asteroid · 01-23-2008, 10:46 PM

I am receving following Error message in /var/adm/messages
My Platform is :

SunOS jump01 5.8 Generic_108528-27 sun4u sparc SUNW,Sun-Blade-100

Code:

"NOTICE: alloc: /: file system full"

Disk space usage is as below:

df -k
$ Filesystem kbytes used avail capacity Mounted on
/dev/md/dsk/d10 76678257 56962561 18948914 76% /
/proc 0 0 0 0% /proc
fd 0 0 0 0% /dev/fd
mnttab 0 0 0 0% /etc/mnttab
swap 547624 24 547600 1% /var/run
swap 547656 56 547600 1% /tmp

Any help ???

paulsm4 · 01-23-2008, 11:11 PM

Hi -

Whatever problem you had with the root ("/") filesystem when the error was logged, it looks like you have 18.9GB free on it now.

Just keep an eye on it...
... and see if you've got any apps out there that might want to gobble 18GB at a time.

Like, for example, the app that should have named itself in the error log.

'Hope that helps .. PSM

Asteroid · 01-24-2008, 02:39 AM

The Error donot point to any process. see below output.
This is what I got in Error.

I guess there isn't any appplication that might be using that much huge disk size.

More over the Error only appears once at 04:2x HRS on daily basis from last one week

Code:

Jan 23 04:30:28 jump01 ufs: [ID 845546 kern.notice] NOTICE: alloc: /: file system full
Jan 24 04:29:22 jump01 ufs: [ID 845546 kern.notice] NOTICE: alloc: /: file system full

paulsm4 · 01-24-2008, 10:13 AM

Take a look at these threads:
http://unix.derkeiler.com/Newsgroups...5-01/2611.html
<= inodes problem?

http://www.sunmanagers.org/pipermail...ne/007355.html
<= file deleted while in-use

Definitely look at the variants of "df", definitely get "lsof" ... and definitely check for any daemons, cron jobs, or monitor apps that might be active on your system.

'Hope that helps .. PSM

chrism01 · 01-24-2008, 05:41 PM

Well, 04:xx hrs is a classic time for people to kick-off housekeeping tasks as there's unlikely to be much real usage at that time.
Some of these tasks can use a lot of temp space during eg backup/compression operations.
I'd start looking at ALL crontabs for jobs starting between say 04:00 and 04:30 (given your posts).

paulsm4 · 01-24-2008, 11:37 PM

Here's another troubleshooting tip, "DTrace" for Solaris:

http://jeffhigham.blogspot.com/2007/...r-solaris.html

And please bear in mind: although the message says "disk full", this might be MISLEADING. The actual problem might be any/all of:

a) No more free blocks (just like the message says)
b) No more free inodes (e.g. too *many* files)
c) User quota exhausted
d) Something else entirely...

'Hope that helps .. PSM