Saturday, September 27, 2008

Dell PowerEdge R900 freezes / locks with Windows Server 2008

We have 3 Dell PowerEdge R900 running Windows Server 2008 they started to freeze at random (it did not blue screen). We have sent Dell and Microsoft a memory dump use the Non-Maskable Interrupt (NMI) on the front of the server.

Microsoft came back saying that the megasas.sys is the problem, this driver is used for the Dell PERC 6/i to run the local hard disk in the server (we had a RAID 10 layout). The PERC is made for Dell by LSI Logic Corporation.

I have asked Microsoft to remove the Dell R900 for Windows 2008 from the Windows Server catalog.

We have spent many weeks working with Dell to fix this problem but they still have not come back with a fix. LSI has to rework the drivers. In the mean time Dell said to change the control but we have reinstalled under Windows Server 2003 and the servers are looking happy.


Anonymous said...

Microsoft really said

. 0 64-bit Kernel summary dump:

There are 8 threads waiting on NtfsNonCachedIo.
One of these threads is processing LfsFlushToLsnPriv and has a LfsLock.
There are 2 other threads which are waiting at LfsFlushToLsnPriv for the thread above to release the LfsLock.
Most of these threads are the system worker threads.

The IOs of all the threads waiting on NtfsNonCachedIo are waiting for IO completion from megasas driver.

The IRPs are:

Irp is active with 2 stacks 2 is current (= 0xfffffa803e21a9c8)
Mdl=fffffa803df2a860: No System Buffer: Thread 00000000: Irp stack trace.
cmd flg cl Device File Completion-Context

>[ f, 0] 10 e1 fffffa803da011d0 00000000 fffffa60011b8590-fffffa803e22ce70 Success Error Cancel pending
\Driver\megasas CLASSPNP!TransferPktComplete
Args: fffffa803e22cf90 00000000 00000000 fffffa803da01320


10: kd> lmvm megasas
start end module name
fffffa60`00ce4000 fffffa60`00cf0000 megasas
Loaded symbol image file: megasas.sys
Image path: \SystemRoot\system32\drivers\megasas.sys
Image name: megasas.sys
Timestamp: Sat May 26 03:51:12 2007 (46576158)
CheckSum: 00012388
ImageSize: 0000C000
Translations: 0000.04b0 0000.04e0 0409.04b0 0409.04e0

Anonymous said...

I have gone through your posts and they are very informative. Unfortunately I could get a few answers I was looking for.
Our company is currently implementing AX 2009 with SQL Server for. In terms of storage we are evaluating EqualLogic. We have planned to virtualize our setup using VMWare. Is it possible to share your recommendation and details of your environment? We are looking for some references and confidence in terms of this iSCSI solution, IOPs, overall performance, if you have a DR setup etc. It will be great help if we could e-mail/discuss over phone.

CustomerX said...

I can, how would you like me to message you?