Lost iSCSI on XEN server after latest update and new Kernel

Last weekend I decided to upgrade a XEN server with its 5 vm’s. Unfortunately, after this action, I lost nss on an OES vm, and it transpired that the Novell Solution was to back rev the kernel due to a c-ck-p by themselves on the VM kernel. After putting this right, I then noticed that iSCSI was no longer functioning. The target computer is the XEN server, and the OES vm connects to 4TB of iSCSI disc on this XEN server. After a good deal of time wasted trying to fix the problem, I noticed that another Standalone backup server was also not connecting to a third 2tb iSCSI disc on this XEN server (the initiator being in this case a Windows 2008 r2 backup server). This pointed to the iSCSI not functioning correctly on the XEN machine after it was upgraded with the new latest XEN kernel. I then uninstalled the iSCSI software and re-installed it to hopefully correct any problems which may have appeared during the patch update cycle. This time, however, there appeared scripting problems with insserv during the install, although the administration console was now available under YAST.

If I tried to make an iSCSI connection to this server from another machine, I could establish a “connection”, but not access any data, and get 100% utilization on the DOM0 of the XEN server. A short term solution has been to uninstall iSCSI on the XEN server, so all the vm’s run at normal speed, and hope anyone here has a solution to resolving this issue. I have tried the re-installtion of iSCSI on the XEN target computer 4 times with the same insserv errors.

Hope someone has any pointers to get things working again.



Forgot to say server is SLES11 SP2

Solved the problem:- upgraded the XEN server from SLES11 SP2 to SP3. The upgrade worked without any problems, and this version has an alternative ISCSI Lio available which is kernel based. Thought I would give this a try first, but even though the initiator could see the drives, I could not find a way to get a successful login and attachment to any of the drives. I tried this from Linux initiators as well as from windows 2008R2 with the same result.

Option two was to disable the Lio version, and try the original iSCSI setup and Lo and Behold, everything went back to normal. Could log into all drives, mount them and successfully get all the mappings back.

Note:- downside of this solution was to lose Backup Exec Ralus on the XEN server, as it is not as yet supported on SP3, and does not work. This often happens with backup exec when a service pack is released.

Hi ChasR,

I just wanted to thank you for spending your time on reporting back your solution, although no-one ever responded to your question. Great!