UNIX for Advanced & Expert Users

Hi,

I got stuck in a very messy situation yesterday after attempting to resolve a filesystem full issue. The user reported that their filesystem (/var/DWS) was 100% full even after moving few files of sizes 14G, 30G and 50G out of the directory. I checked the filesystem once more after this complaint, and realized that there was a sub-directory named "compressed" within this directory that was the largest within it. I asked them to housekeep it, but they said this directory size was only 6.7G, which was way smaller in size than the total files/directories they moved out of /var/DWS. The filesystem was an ufs mount.

So I suggested to un-mount and re-mount the directory, only to get into more problems, as the person who gave approval to do this, did not properly stop the applications/processes that were accessing this directory. So I could unmount the directory (umount -f /var/DWS, as a normal umount did not work). But after this I could not re-mount the directory, as it throws error "Device busy".

So, I rebooted the server, and then the server could not boot up. It got stuck in a blank grey screen (Like a tv channel that does not have a transmission).

Please throw some light on how to start troubleshooting this issue? Better if there is a solution for this issue?

It would be a good idea to tell us all what the hardware is and especially what the operating system is!!! Otherwise we're all left guessing.

The server is a SUN SPARC T3-2.
And the Operating System is Solaris 10.

So when you power up the first thing you should see is the service processor. That will go through diagnostics and then present you with a prompt (unless it's configured to automatically boot).

So what is the last thing you see before the screen goes blank?

Do you get a service processor prompt eg, '>ok'?

Hi,

It does not reach the ok prompt :

Code:

-> start /SP/console
Are you sure you want to start /SP/console (y/n)? y
Console session already in use. [view mode]

Serial console started.  To stop, type #.

Yes, because you didn't remove all power, ie, all power cables disconnected.

The service processor is talking to the main system (trying to talk to Solaris as console). Do what is says and type

Code:

#.

(That's a hash keystroke followed by a period keystroke. No 'return' needed.)

And you should get the service processor prompt. You might need to bash the return key a couple of times to wake it up.

You can then issue SC processor reconfigure commands or, more likely, a 'boot' command.

If it fails to boot normally, try putting your Solaris distribution media into the DVD drive and boot from that into single user:

Code:

 
 ok> boot cdrom -s

That will tell you whether the hardware is capable of running Solaris or you have a hardware fault.