VAR full

[Mike Rightmire]

I am running HPUX 10.20 and the VAR is filling up. The wierd part is the bdf shows the /var as having 1 GB (which is correct) but using du -sk on /var shows only 9 megs of data. Where has the rest of the space gone.

We are running rogue wave, which I wonder if its allocating space...?

Any advice greatly appreciated!
Mike

[Vincenzo Restuccia]

lvdisplay -v /dev/vg00/lvol8,
vgdisplay -v vg00 (check free PE,1PE=4M).

[Bill McNAMARA_1]

what does lvdisplay /dev/vg00/lvol8
show.

In my case:
LV Size=616M

bdf
614400

du -sk /var
15583

which is around the space used in my /var (from bdf)

Bill

[Mike Rightmire]

Hey Vinc,

Thanks for the quick reply. I did the two checks and (using the lvdisplay) it shows 250 avail and 250 alocated with everything as current (???) and with vgdisplay it shows total PE = 2168, allocated=1269 and free=899.

What exactly am I looking for?

Thanks,
Mike

[Vincenzo Restuccia]

It is correct,the problem should be with du command.Verify the integrity of data with:
cd /var
du -k|more and find /var -size +100.

[Bill McNAMARA_1]

can you post up the output of the commands.

Your lvol looks fine 250 current (not requiring mirror synchronisation or failed disk bits) Physical extents.. 1GB lvol size.

bdf should show under the kbytes column lvol Size * 1024

du will show the size used, and should be related back to bdf used column not to the kbytes column.

Later,
Bill

[Bill McNAMARA_1]

mike, correct me if I'm wrong, but are you getting mixed up between the df and du command!

Later,
Bill

[Bill McNAMARA_1]

BTW, only if you have onlineJFS can apps preallocate space, to stop defrag effectively, using a setext command.
If you have OJFS, use the getext command to find for preallocation.

Later,
Bill

[John Bolene]

do this
cd /var
du -x|sort -rn|more

this should show the culprits using the most space

I would imagine that you may have some crash data in /var/adm/crash, some syslog stuff in /var/adm/syslog, /var/tmp may have some stuff, print goes in /var/spool, cron logs go in /var/spool/cron

[Mike Rightmire]

Hey!

Here is the information from some of the recommended commands. I am not confusing df and du per say...what is confusing me is that bdf is saying I have 1 GB of space for /var and it is full; du is saying there is only 22 megs (now) of data in /var.

So what is taking up the other 978 megs of space?

The "df -b" command produced ...
/var (/dev/vg00/lvol8 ) : 106 Kbytes free

The "bdf" command produced ...
Filesystem kbytes used avail %used Mounted on
/dev/vg00/lvol 1024000 1024000 0 100% /var

The "du -sk /var" command produced ...
22444

The "find /var -size +100" command produced ...
/var/opt/ignite/local/install.log
/var/opt/perf/metdesc
/var/opt/perf/rxitemid
/var/opt/perf/status.scope
/var/opt/perf/status.alarmgen
/var/adm/syslog/syslog.log
/var/adm/btmp
/var/adm/qhap.log
/var/mail/root
/var/sam/log/samlog
/var/sam/log/samlog.old
/var/sam/boot.config
/var/stm/config/tools/diagnose/summit_crc_b
/var/stm/config/tools/diagnose/summit_crc_c
/var/stm/config/tools/verify/motestdata
/var/stm/config/tools/verify/summit_crc_b
/var/stm/config/tools/verify/summit_crc_c
/var/stm/data/uut_status
/var/stm/logs/sys/activity_log
/var/stm/logs/sys/diaglogd_activity_log

Hope this helps!
Mike

[melvyn burnard · Accepted solution]

hmm, you may have had a file opened by a process which has been removed while it was still open.
The normal way to cure this is find the process using something lik efuser, or reboot the system and see if that frees up the space.
If it does, monitor var closely and try to discover which process is writing to it.

[Pedro Sousa]

it's even more strange than that, because at the begining of this post you had only 9MB with du, and now your higher than 22MB!!
Check if you have a backup of /var before you execute a shutdown!

[Mike Rightmire]

Hey folks,

Thanks for all the help and great advice. As it turns out, the problem was a process which was still running (Qualex), but was having some difficulty.

I used fuser -cu /var to see what was touching the drive and started examining and removing the processes one by one. When I stopped the qhap process (Qualex) my var went from 100% used to 4% used.

I restarted the machine, including Qualex, and the var dir has stayed where it should be. Now I just have to watch Qualex to see what was going on!

Thanks again!