Well in Episode 3 of our saga Many goes bald (again) and (almost) drop kicks the server out the window. After going through the pain of installing 10.1 and updating everything I proceeded to upgrade to OpenSuSE 10.3. The upgrade went okay as far as the boot DVD was concerned (no errors during boot), but the dreaded SCSI errors from Episode 1 came back when I finished the upgrade and rebooted the machine. After about 10 minutes of repeating the F-word to the machine I decided to check on the net again, this time not at google.com, but at their linux centric search page google.com/linux. Well I have to admit, google is really god :-). Someone there had suggested an upgrade to the Adaptec PERC 3Di raid scsi card that comes standard with the PowerEdge 2650 and guess what, that solved everything. I’m not gonna bore you with the details of my 90 minute hunt for 3.5″ floppies around work to put the disk images on. Let’s just say I’ll be glad when someone invents the “Network Upgradeable Firmware”…..aka ‘NUF :-)
So it looks like there are two ways of getting the latest linux kernels to barf when it sees this type of configuration (Dell 2650, PERC 3Di scsi raid card). Installing a “XENified” kernel and/or not having the most up to date firmware. The symptom is excessive error outputs from aacraid driver.aacraid: Host adapter abort request (0,1,1,0)
aacraid: Host adapter abort request (0,1,1,0)
aacraid: Host adapter reset request. SCSI hang ?
aacraid: Host adapter abort request (0,1,1,0)
scsi 0:1:1:0: scsi: Device offlined - not ready after error recovery
The cure is to find yourself 3 floppies and upgrade and don’t install XEN. The known working config is BIOS version A21 and PERC 3Di firmware 2.8-1[7692]. Hopefully this is not going to be the last of it. THE END….FIN. :-)