Conference decwet::ntaxp

1024.0. "System hangs when a DR Watson log is written" by NQOS01::alfras1_port38.alf.dec.com::WORKBENCH () Wed Apr 02 1997 16:26

We have seen this problem on two seperate Alpha 4000's running NT V3.51 SP 5

If DR Watson appears and starts writing the log file and you ALT+TAB and 
start some other application then the system will slow down and hang. But if 
DR Watson starts writing the log and you wait until the log file has 
finished, click on OK and then start something else then the system doesn't 
hang. 

The customer that I am at is having a fit.  Because this are production 
systems.  On one system Dr Watson started when the user started Notepad.  The 
other system Dr Watson started when NetWorker Administrator started and 
NetWorker was stopped.

We sent the first drwatsn32.log file to the CSC and they said the problem was 
a memory access violation so couldn't pin point the problem.

Well the other day on another Alpha 4100 running NT 3.51 SP5 when the backup 
(NetWorker Save & Restore) was running, a DR Watson started when NetWorker 
found a bad tape and the backup stopped.  When I was checking how the backup 
ran the next morning, I started the NetWorker Administrator Program and since 
NetWorker stopped during the night I also got a Dr Watson dump and changed 
the screen to look at the event logs and when I realized that Dr Watson had 
started the system started to slow down and then hung.  I have both Dr Watson 
log files and will send them to the CSC to review.

But in the mean time, this seems to be a problem with NT.  Have you seen this 
problem before? and what can I do to get this fixed?

Both systems are:

Alpha 4100  (21161 299MHZ)
Alpha Bios 5.12
NT 3.51 Service Pack 5
Memory 655,336KB
Both have a KZPSA adapters and a fault tolerate (Raid) drives.
Also have another SCSI adapter that goes to a jukebox for tape backup.

The systems are setup to implement NT Clustering but not using the Clustering 
today.  

Thanks in advance,
Allen Curle

Please let me know what other information you might need.

We are going to try updating the firmwarm and HAL code to see if this fixes 
the problem.

We are trying to duplicate this problem.  We created a file (called BadApp) 
and it calls memory area 0 which will crash and start Dr. Watson.  Can 
someone create a badapp for us and compile it under the Alpha so we can 
trouble shoot the problem? If you need it, I can include our source code.

Thanks in advanced,
Allen Curle
Curle@mail.dec.com

There is a known problem on NT (alpha only) with Dr. Watson getting in an
infinite loop. I am hoping the fix from MS gets in the upcoming SP3 release
but it doesn't look good at the moment.

If your system is truly "hanging" rather than just slow because of Dr. Watson
sucking up all the CPU time then you have a different problem.

This is exactly the problem we are having.  The system will slow down and 
when I said hang I meant the system was so slow it seemed like it was 
stopped.  Is there anything I can do to pressure Microsoft? (I thought you 
needed a good laugh)

The customer I am working with is Philip Morris which also owns Miller 
Brewing and Kraft Foods.  We are slowly getting the Alpha's in the door and 
replacing Sun systems, IBM systems and Compaq servers. I didn't know if that 
would help.  I know Microsoft couldn't care.

Thanks for your help,
Allen Curle
Curle@mail.dec.com

Since you said SP3, I am assuming that the problem existing with NT V3.51  
still exist with NT V4.0?

I need to be carefull with the customer since they are looking to find 
anything to not buy Alpha's and to stay with Compaq.

Does the problem still exist with NT V4.0?

Thanks in advance,
Allen

The problem still exists with WNT 4.0, both with SP1 and SP2.

Now the bad news. On April 11 the Dr Watson bug fix was postponed to SP4. 

Believe it or not, it does help to have real customers who want this fixed
rather than just us engineers.