Hi!
We have two Dell R610 servers containing:
SAS 15K RAID-1 volumes
Hyper-V
Clustering
Starwind SAN - the latest v.
OS: Win 2008R2 Ent.
We have Intel ET 1Gb NIC (dual port) where:
1. port is connected via xover cable to the other server (host) same port - for StarWind syncing
2. port is connected via xover cable to the other host same port - for ISCSI targets
SAN's targets are all configured as HA and man... all the tweaking we did to get that sync to work faster has been in vain. We have reviewed all your blogs and posts available on the subject and related and tried all the suggestions within our scope of possibilities here but sync won't budge past 25% of line utilization. with ntttcp we can get 800Mbps going through it and disc write speeds are close to 200MB/s on all the RAID volumes.
But anyway,, this isn't the topic for that.
The struggle here with ISCSI has been to keep the targets going through the right NICs.
Your wizard clearly asks what adapters to use for the sync - and that should be that.
However, if a node is rebooted, lets say, you will notice the ISCSI Initiator adding another portal to your configured one and that newly discovered, automatically added portal is the other partners sync adapter.... it looks like then your MPIO ends up using that adapter also for ISCSI target path distribution.
Does it not?
Any idea how to prevent this from happening?
Also, any better solution to restoring HA if both nodes had to go down (such as a prolonged power outage)?
Thank you kindly for your help,
Darko
The Latest Gartner® Magic Quadrant™Hyperconverged Infrastructure Software