3

I have 3 blade servers that are Blue Screening with a 0xC2 error as far as we can tell randomly. When it started happening I found that the servers weren't set to do provide a dump because they each have 16GB RAM and a 16GB swap file divided over 4 partitions in 4GB files.

I set them to provided a small dump file (64K mini dump), but the dump files aren't being written. On start up the server event log is reporting both Event ID 45 "The system could not sucessfully load the crash dump driver." and Event ID 49 "Configuring the Page file for crash dump failed. Make sure there is a page file on the boot partition and that is large enough to contain all physical memory."

I understanding is the the small dump shouldn't need a swap file large enough for all physical memory, but the error seems to point to this not being the case. The issue of course is that the max swap file size is 4GB, so this seems to be impossible.

Can anyone point me where to go from here?

Randy K
  • 75
  • 1
  • 9
  • For what it's worth, the blade servers don't have physical disks in them. They boot off of an HP EVA440 SAN. I'm starting to think that maybe the issue here. – Randy K Feb 24 '11 at 21:17

1 Answers1

1

Problem is due to booting the servers from the SAN.

Microsoft white paper on booting Windows Server 2003/2008 from a SAN

Seems while Windows 2008 is able to write a dump file to the SAN boot drive, Windows 2003 isn't.

Randy K
  • 75
  • 1
  • 9
  • As an update, after contacting Microsoft, they provided a hot fix for the STORPORT.SYS driver that allowed a dump file to written to a SAN from Windows 2003, allowing us to track down the original issue causing the BSOD. – Randy K Apr 05 '11 at 21:19