Strange problem on New DL580G2...

[MARTIN_67]

Hi

I am facing a strange problem on all our DL580G2

Here is the situation:
We are running SAP ( DB and CI for those who know SAP) on a DL760 (4Gb and 7 procs)with windows NT

For the users we have 4 Applications Servers which are networked with the database (on the DL760)

So these Application Servers (DL580G2 5Gb and 3 Procs) are installed with W2kAdServer US .

The problem we have is that suddenly one of the AS stops running SAP (but still responds to ping ) and when we connect (via ILO) the server is still UP but we cannot do anything , the screen is there seems normal the mouse works but we cannot start anything....)

The only solution is to reboot and restart SAP

This append on every servers at least one time !!!

The 4 AS were functionning well for 10 days and then suddenly one "freeze" sunday and another one "freeze" today !!!!

We have no logs on the event viewer when this appends !!
and moreover we see no explanation of this problem !

we try to put one of this server on NT 4 to see if this is related on the OS ...

So if someone has any idea about this behaviour (hardware , software..)
please help us

Thanks

[Stuart Robinson]

We've experienced the same problem. Check out http://www.intel.com/support/platform/ht/os.htm. Intel say HyperThreading is not supported by them. We're disabling our Hyperthreading and waiting to see what happens.

[Stuart Robinson]

Disabling hyperthreading has had no joy. Our box is still unstable. We are now looking at a potential video driver issue.

In the SSD only ATI Rage IIC PCI drivers are supplied, but on DL580-G2 server the video card is an ATI Rage XL PCI. I have spoken to other contacts and ATI write a specific fix for the issue.

Current driver version is 5.0.2184.1 and the version my contacts have is 5.0.2195.4015.

[Stuart Robinson · Accepted solution]

Check out http://wwss1pro.compaq.com/support/reference_library/viewdocument.asp?source=EL020923_CW02.xml&dt=3&pid=-1 as well

[MARTIN_67]

Hi

I was out of the office the last two days... we also had disabled the HT on two of our servers to see what happend , as for now they are still ok.

But the last thing you say is very interesting, the serial number of our servers "says" that they have been produced the week 37 !! so they could possibly have the first criteria for the VRM problem , we will look in the data center to see if they have the second one.

Thanks for your help

More later

[Jeff Mathews]

I just wanted to note that the article by Intel reference above is for the Hyperthreading of P4 chips, and not the Xeon model.

Jeff

[Jakub Nowakowski]

We have similar problem with DL580. We have 3 x 1.6 Xeon / 6GB RAM / Win2K Advanced Server. After the console hangs we can ping the server. This server was produced in 46th week. We suspected HBA card/drivers, but it was a wrong trace. We found newer ATI drivers for the video card 5.00.2179.1 and 5.0.2195.5012. Can somebody comment this or recommend other solution. TIA.

[Jakub Nowakowski]

Ha!

It seems we solved our problem! There is a SQL Sever Enterprise installed on this server. For the testing purposes it is configured to use fixed memory size. The fixed size is pointing to the maximum possible size! Some additional applications (we used pinball to test the video adapter driver) are causing the problem. We changed for a while the SQL memory parameter to smaller fixed size. And we started dozens of applications. Everything worked O.K. Now we are testing this server with VA 7100, 2 HBAs and AutoPath installed. Nothing freezes!!!

[Stuart Robinson]

We have changed out servers to boot using VGA mode (/basevideo). Since doing this, the boxes have been stable.

Out ATI Rage XL driver is that which ships with MS Windows 2000 server, version 5.0.2184.1 dated 18/11/1999.

We have been advised to apply the Support Paq after any Windows service packs, but if the video driver is the issue, it will never get updated as it's not included in any Support Paq.

We have raised this issue, but have been advised that the updated driver isn't shipped in the Support Paq as the Microsoft driver is the correct one.

We have no plans to implement newer drivers from ATI though.

[Stuart Robinson]

I have discussed the SQL fixed memory issue with our DBAs and they have explained the following: -

By configuring the fixed memory to the maximum, the operating system cannot get memory back from SQL to run other applications, hence the server "hanging" when trying to run additional applications. They recommend that if you fix the memory, always leave yourself about 20% for the operating system to perform additional processes, and where possible don't have any other applications running on the server.