Ghosts and Goblins in the Server 1

[BKtechie]

Hello all-
I have a Novell 4.11 Server that routinely disables network access around midnight every night. It does not report anything in the log from the powershoot software or any hardware errors. After researching the Novell site, the closest solution I saw was to replace the network card. This did not achieve anything. I have 29 other servers with the same hardware configuration that do not crash. Does anyone have any experience with this sort of behavior? Thanks in advance.

BKtechie

 

[TheLad]

Are you running any backup software on the server that is configured to disable login? Or is it properly abending? If it is abending, it will be logged in SYS:SYSTEM\ABEND.LOG.

Have you got CONLOG configured and running? If so, check out the console log file as well. -----------------------------------------------------
"It's true, its damn true!"
-----------------------------------------------------

 

[salserocito]

Are we sure that the server is even crashing? Or is it just disconnecting Users? Does the Server show an Uptime of only a few hours, or a few days?

If the Users are being disconnected, have you seen any Watchdog messages?

TheLad is right (...as always! )about CONLOG... if there is anything happening on the Console, CONLOG will show it.


Good Luck!

 

[BKtechie]

Which is the CONLOG? I have the Server Console Log sent to me daily through some scripts. It sends any messages from the past month above Severity 4. Oddly though, I only got the message that it Remirrored the partitions this morning, but nothing about dismounting volumes, a system shutdown, or users getting kicked off of the system. The abend log was clean as a whistle with the last entry sometime in yr 2000. Just for good measure I checked out all the other logs in Sys:System\, but returned empty. Apparently they had to reboot it this morning, so I was told, and this was reflected in the in the UPS software log. However, all the other re-boots "they tell me about" are not appearing. Anywhere else I could check? Thanks for your input.
BKtechie

 

[TheLad]

CONLOG.NLM logs all events that are viewed on the system console to a text file called CONSOLE.LOG

Add the following line to the AUTOEXEC.NCF (and load it manually on the console) to get it to work:

_______________________________________________________
LOAD CONLOG FILE=SYS:SYSTEM\CONSOLE.LOG SAVE=SYS:SYSTEM\CONSOLE.OLD MAXIMUM=500 ENTIRE=YES
_______________________________________________________


Note: this is all on one line. After a reboot, the old console log gets saved as SYS:SYSTEM\CONSOLE.OLD so if the server was to reboot you could check this file to see what happened. -----------------------------------------------------
"It's true, its damn true!"
-----------------------------------------------------

 

[BKtechie]

Thanks for the info on CONLOG. It will come in handy in the future. The good news is I have more info about what's going wrong on the server now. The bad news is that the server stops durring the loading process. Someone attempted to reboot the server before I came in. Apparently one of the NLM's is turning off low priority threads. It provides the message to Set Upgrade Low Priority Threads = ON. Unfortunately I can't get to a prompt to attempt this command. Is there a log of the batch file that loads the server so I can attempt to track down the NLM causing the problem? Thanks for all your help thus far.

BKtechie

 

[TheLad]

You can look in C:\NWSERVER\STARTUP.NCF or SYS:SYSTEM\AUTOEXEC.NCF. These are the two main files. They may call other NCF files so you will have to check these out as well (an NCF file is the NetWare equivalent of a DOS batch file - in laymens terms!) -----------------------------------------------------
"It's true, its damn true!"
-----------------------------------------------------

 

[BKtechie]

I've checked the batch files on a working server. How can I get into the SYS:SYSTEM directory on the non-working server through a DOS boot disk? If I can edit the autoexec.ncf there, I can remark out some of the NLM's to get into the server. We can't get past a message that an NLM has turned off low priority threads. I can't even get a prompt. Thanks for everything.
BKtechie

 

[TheLad]

Best thing is to boot the server up into DOS and run: SERVER -NA

This will boot the server up and mount SYS, but it will not run the AUTOEXEC.NCF (which is probably where the problem lies). Once the server is booted up, you can use either EDIT of NWCONFIG to edit the AUTOEXEC.NCF

EDIT SYS:SYSTEM\AUTOEXEC.NCF

or

NWCONFIG / NCF file options / Edit AUTOEXEC.NCF -----------------------------------------------------
"It's true, its damn true!"
-----------------------------------------------------

 

[Jaapvs]

the message to Set Upgrade Low Priority Threads = ON means that the server is trying to load a .NLM that will not work as long as this setting is set to OFF (the default setting). Probably some third party software.
Best thing to do is to put the line "Set Upgrade Low Priority Threads = ON" in the autoexec.ncf, to stop the server from crashing.
You can also find this setting in servman under misscelaneous.
It might be wise to look for patches for this (third party) software as this new setting may slow your server down a bit.

 

[BKtechie]

That line already exists in the AUTOEXEC.NCF. The software is apparently turning this off, or loaded before that line in the AUTOEXEC that says "set Upgrade Low Priority Threads = ON". I am going to drive to the site to see what is going on there. Thanks for all your help,
BKtechie

 

[yvfwc]

Another handy switch used when loading Conlog that TheLad showed above is: (at the end, after entire=yes) ARCHIVE=YES. Instead of only keeping the last version of conlog, it will auto number conlog files (conlog.001, conlog.002, etc) so if there is multiple abends or restarts before checking the conlog file the history will start to build. It has been very useful for me and the file size is small so disk space is not an issue.

Cheers,

Mark ;-)