iSCSI Targets in HA Used By OS X Desktops Directly

[Ciscogeek 0]

Hey SNS,

I had posted this on StarWind's forums, as I think this is an issue for them, but could you a look and see if this could be an issue on GlobalSAN/iSANmp's side?

 

https://forums.starwindsoftware.com/viewtopic.php?f=5&t=4577

 

Thanks!

[SNS Hal 1]

Hi,

If we're understanding correctly what you're after, creating an AppleRAID (RAID 1) for the two targets should take care of it.

In this case, SANmp (iSANmp) would only see one volume, and continue writing without trouble in the event one of the targets went down. The software RAID could then be rebuilt once access was restored. Hope that helps - feel free to open a support case if we can be of further assistance.

[Ciscogeek 0]

Hey Hal,

There's a misunderstanding here for sure.  We should not have 2 disks.  The Starwind (SW) software manages the HA disk between the hosts and related replication.  The targets that appear to Globalsan are the exact same disk from both servers.  When you mount one server's target, you can see the RW state of the other disk immediately update as well.

This is something of a potential multipath issue I think.

[SNS Hal 1]

I thought AppleRAID may not be what you were after, but could be an alternative solution.. Multipathing is handled by the operating system, rather than the initiator. The OS can correctly collapse targets if the same target is seen over multiple paths, but in this case the OS has no awareness that there is actually only one intended target. We expect if you can get the Starwind software to make each LUN appear identical to the OS, that you're correct that multipathing could be the solution.