NW 6.5 SP1 Server hangs

[AdmREA]

We've two NW 6.5 SP1 Servers (CPQ ML370 G3 and CPQ DL360 G3, NIC Driver N1000)here and both have the same Problem: from time to time, I lose the connection to them (can't ping, remote console won't work). At the server console, anything looks normal, I can switch through the Screens, Util is somewhere around 10% - 20%. I can press Enter in the command screen and a new line in the console appears, BUT: when I enter a command (for example a ping) the console hangs up. I can still switch through the screens but when I choose an option in the monitor, the monitor also hangs up (switching still possible). Pressing CTRL-ALT-ESC brings up the Emergency-console, when I choose Down the Server nothing happens. I can just restarting the server by going into the Debug-Screen and then exit to DOS.

Does anyone have similar problems with NW 6.5?

 

[marvhuffaker]

Sounds very similar to a problem I had recently. Loading DOSFAT.NSS resolved it for me. You can do a groups.google.com search for "CDBE GREMLIN" and find many references to it with explanations. If you could get into Monitor while the problem occurs, you should try to see what is hitting the CPU the hardest. I bet your utilization is higher than 20%, but monitor is not displaying correctly.

Marvin Huffaker MCNE, CNE
Marvin Huffaker Consulting

http://www.redjuju.com

 

[1stITMAN]

Plz check all your logs and post errors. Thanks

Zaheer Ahmed Iqbal
I.T Systems Support Engineer
Bsc. (Hons).

 

[AdmREA]

Unfortunately there are no Errors or messages of interest in the Logfiles. The End of the Logfiles is the last entered command.

 

[marvhuffaker]

Admrea, i coudn't get log files when I had that problem either. Try putting this in the STARTUP.NCF file..

set auto restart after abend=0
load DOSFAT.NSS
set auto restart after abend=1

Then restart your server and see if you still have the problem.

Marvin Huffaker MCNE, CNE
Marvin Huffaker Consulting

http://www.redjuju.com

 

[ITsmyfault]

Marvin - why do you set the auto-restart back to 1? I thought this was a no no for servers where dosfat is loaded? I'm running clusters here so it's always 0..
Just curious. Thx for the heads up on that. Found a good link on it here:

http://lists.indstate.edu/guest/archives/oper-l/log0310/doc00000.doc

very interesting..

 

[marvhuffaker]

Setting Auto restart to 0 first allows DOSFAT to load without hassling you to confirm that you really want to load it. It's the only way to load DOSFAT automatically in your startup files.

As far as setting it back to 1, that is a good topic for debate. If you understand the "Auto Restart After Abend" settings, you know that when it is set to 0, the server will drop to the debugger and wait for your response any time it abends. If it's set to 1, it will try to recover and continue to run.

Now, given this, which is worse? Possiblity of corrupting the DOS partition (Option 1) or corrupting 3000 files that are open by users when the system abends, halts, and goes to the debugger (Option 0), essentially dropping all communication in and out.

I try to only set it to 0 when I am specifically trying to capture a coredump to send to Novell for analysis. Of course it will depend on the specific environment.

Specific to DOSFAT.NSS, I try not to run this at all, there are other ways to access the C: drive. I've only seen a couple cases where I needed to run DOSFAT permanently to keep the server from hanging due to a problem with Arcserve.

I've also read the TIDS on this forward and backward and don't quite understand why writing an abend.log file can corrupt the FAT table. The TID says that it writes info to the abend.log file located on DOS, but it's actually written to SYS:SYSTEM\abend.log. SO i'm a little confused on that topic.




Marvin Huffaker MCNE, CNE
Marvin Huffaker Consulting

http://www.redjuju.com