Print server randomly crashing 1

[animal74]

I have an NT 4.0 print server that has been randomly "crashing". It approximately happens once a month at any time of day (usually the most inopportune times!). When I find it - the machine is on but the screen is black. A manual shutdown/reboot solves it and it runs fine until next time. The Event viewer just says "the shut down was unexpected". There are no errors that are visible. Could it possibly be over-heating? The room is AC'ed. Any suggestions would be appreciated.

animal74

 

[RichardMoran]

Hi animal74,

When servers randomly crash it can be difficult to establish the cause unless you have a good understanding of interpreting any results you get from memory dumps/event logs etc... Even then it may only point you in the right direction, you still may not be able to fix it yourself.

If by 'once a month' you mean that the average interval between crashes is a month, the crash could be caused by a cumulative fault. For example, some applications/drivers are poorly written and cause problems such as not releasing memory when they have finished with it. Over a period of time your available memory could become less and less, until ultimately there is none left. At this point a crash could occur.

If it is this sort of problem, you may need to do some research on your printer drivers to see if there are known problems, and if later versions have corrected it. Until then, you need to protect yourself against the server crashing at an inopportune moment. You could achieve this by scheduling an automatic reboot of the server when it is not being used, perhaps at the weekend, or even overnight. This should clear any problems, and more importantly gives you back control of when the server reboots.

If you adopt this routine of rebooting your server, and the server no longer crashes you know that it is most likely caused by a cumulative problem. If it still crashes then you need to start looking at other possibilities.

I think that over-heating is unlikely, because as you say the room is AC'ed. If the room is cool, then any excess heat is being removed by the AC. Most servers have a means of monitoring the internal temperature, even if only through the BIOS. If this option is available you can compare the temperature against the manufacturers guidelines to ensure that it is running at a suitable temperature. CPUs generate more heat when they are put under more load. I think a print server is highly unlikely to strain even a modest processor. You can verify this by viewing the CPU activity.

Hope this helps. Good luck.

Rik

 

[animal74]

Thanks Rik,

I was thinking of implementing a reboot schedule, but haven't tried it yet.

Unfortunately, that "inopportune moment" may have already arrived. Today, the server has crashed twice, each time it hangs on a blue screen with a "Stop" error with this message: "IRQL_NOT_LESS_OR_EQUAL".

The error in the Event Viewer is this:
"The server was unable to add the virtual root '/' for
the directory '\\Yzf_7f_server\7fcommon\Marketing' due
to the following error: Logon failure: unknown user
name or bad password. The data is the error code."

I haven't had time to put a backup server together yet.

The server is running SP 6. Thanks for any more suggestions!