Nobody can't logging to this server due to space problem.

[HP UNIX Professionals]

HI,

We have HP 9000/ HPUX Box, nobody can't logging to this server due to file system full. Anybody help he sort out this issue.

We are getting error message like

"msgcnt 94 vxfs: mesg 001: vx_nospace - /dev/root file system full (1 block extent)
Connection to XXXX closed"


Thansks in advance

vinay

[Suraj K Sankari]

Hi,

Reboot the box into single user and do the house keeping job at of / file system.

Suraj

[Robert Salter]

I agree with Suraj, the /dev/root file system is full, so boot into single user, find out what's using the space and clean it up.

[Julián Aimar]

Reboot and login in single user mode, process,

*******************
You should interupt the booting when it ask you to press any key within 10 secs.

It will take you to menu, where you need to give this command.

Menu> bo pri

It will ask, do you want to interact with IPL Y/N ?

Give "Y" . It will take you to ISL prompt. In ISL give this command.

ISL>hpux -is

This will boot the system into single user mode.

*******************
if you work with opsware it's a valid option too


Saludos desde Argentina..!!!

Julián

[Dennis Handly]

Anyone still logged in may be able to remove enough files from /tmp (if in /)?

I hope you learned that an admin needs to monitor file systems for being full.

[Steven E. Protter]

Shalom,

du -sk | sort -rn | more

# Find out where the big stuff is.

Best to run that in single user mode, though you may need to mount /usr to do so.

This is probably something you will need to do at the console.

With most filesystems unmounted, take a look around.

Any files sitting in /var when not mounted take up space when the system is running but can not be deleted. I call this the old hidden file trick.

SEP

[Dennis Handly]

>Any files sitting in /var when not mounted take up space when the system is running but can not be deleted.

Years ago, I was able to export the root filesystem and then delete the files out from under the mount point.

[Ian Dennison_1]

Clever trick to resolve without a reboot(especially if you have a backup utility like Data Protector).

Create a script that does an "ll /" and several "du -sk [directory]" or "ll [file]" on likely candidate directories for the file that has filled up root (maybe a restore or a large logfile). Redirect output to a file on another mountpoint if you want.

Set as executable, back it up using Data Protector and restore to /opt/omni/lbin on the stuck Server (rcp usually required space on / to work, DP does not).

Next, configure a backup to run a pre-exec of the script on the stuck server and to backup 1 or 2 files. Run the backup and note the output of the space interrogation script (either in the logfile or on the backup screen).

Once you have identified the file or directory causing problems, use the same method again to create a script to remove the culprit file / directory - write script, set executable, backup from one server and restore to problem server into /opt/omni/lbin, set as pre-exec for a backup and run the backup.

Have used this several times, its fiddly but works a treat if you can't get approval for a reboot. I always create a 2MB file in root named "/.RESERVED_SPACE" which I use this method for temporary reclaiming space long enough to get logged in.

Share and Enjoy! Ian

[Steven E. Protter]

Yes,

I heartily endorse having a slug file in / that you can get rid of when you are in trouble.

Thats a good thing to do going forward.

If it happens to be an 11.00 system, they have an ugly idiosyncrasy to actually damage /etc/group or /etc/passwd when / is full. Have a backup ready.

SEP