Currently a failover of an ISCSI volume hosting a Hyper-V Cluster Shared Volume will tend to cause the host VMs to blue-screen. This apparently is due to the ISCSI-3 persistent reservation locks being stored in RAM and not replicated.
Is there any news about when this problem will be resolved?
A feedback from developer team I received lately says this issue is on the table right now, but there is no estimated patch release date yet.
Currently it is not advised to use automatic failover with HV cluster.
Fortunately, I was pulled off of that project and never actually implemented it.
Tim, I'm curious, after the VMs blue-screened where you able to restart them successfully?
Well great news that this is on the agenda. It's a critical bit of missing functionality and a shortcoming I didn't see mentioned anywhere in the sales material.
Olias - I have a setup much as you describe in your other thread, a cluster of 4 hyper-v machines with 2 DSS storage boxes relpicating all our VMs to another building for disaster recovery purposes. I set this up about 18 months ago but didn't at that time set up failover as I ran out of time before it had to go into production. I figured at least I had a replica and could do a manual failover should it be needed.
I managed to get some down time on the system before xmas and set up failover. This works in the sense that if the primary DSS box goes down, the secondary takes over, but the side effect seemed to be that some / most of the Hyper-V hosts would blue-screen and the linux hosts would have their file systems switch to read only. All the machines came back ok after a reboot, but I wouldn't like to push my luck with this. I suppose it's equivalent to ripping HDs out of a running server, probably not recommended.
Open-E support helped me with the issue and we went through updating drivers, RAID firmware etc before they eventually told me that it was a known issue - failover didn't work with Hyper-V.
This was kind of disappointing having invested significant time and money in the DSS solution, which I'm otherwise very happy with. The problem is without working failover to do any maintenance / updates etc on the primary DSS box I need to shut down all the virtual hosts, which is most of our infrastructure. Hopefully they will come good with an update soon!
I agree, it is a bit disappointing. Actually, the whole subject of Hyper-V and Open-e is a disappointment. I started using Open-e solutions several years ago as it was/is a very cost effective storage solution. And as Hyper-V is also extremely cost effective (free for Windows users) the two products would seem to be a very good match.
However, there does not seem to be any emphasis from Open-e on Microsoft's virtual servers. For example, searching the Webcasts and Videos section on the Open-e web site lists twelve titles with "VMware", three with "Xen" and zero with "Hyper-V" in the title.
Don't get me wrong, I'm not bashing DSS, it's a fantastic product; I just wish there was more effort put in to Hyper-V support/functionality.