I have an msa1000 hosting the storage for a novell cluster. The LUNs are presented to the Novell which via the media pool assigns space to Novell Volumes. As far as I understand Novell should be completely ignorant of teh inner workings of the SAN, only being aware of teh LUNs presented to it. However the last two times I have had a drive failure it has resulted in the Novell cluster losing temporary visibilty of its volumes. Also when the replacement drive rebuild is completed, again the Novell Cluster loses temporary visibilty of its volumes. I am running debug which shows nothing. Does anyone know why this could be happening? thx
As an addition to this discussion... I have noticed that after a disk failure, the on-line spare rebuilds fine. Once this has completed I replace the original drive and this then rebuilds and the on-line spare returns to being an on-line spare. Surely this is incorrect and the replaced drive should just become the new on-line spare?
I have now been advised that setting rebuild priority to High can mean a minor LUN interupt when rebuild kicks in and stops. Novell 6 clustering is too fragile to handle this and so volumes drop out. I have therefore set rebuild priority to Medium to get around this problem. If anyone could answer the point made in my 2nd reply it would be appreciated. Thx
Re-your "2nd Reply", this is exactly how the SMART technology works (the MSA1000 /MSA1500cs) is a SA 5304 after all), so YES the original spare will return to being just that. The idea being that the original disk positions forming the array will be maintained, so previous configuration for performance, and availability will not be affected. Seasons Greetings from everyone at HOST, and I look forward with interest to more of your postings in the coming year.
