ISCSI config question with StarWind VS with Hyper-V

[datastream187]

Hi all

Im looking for a little clarity about the ISCSI settings that are given in the "StarWind Virtual SAN Hyper-converged 2 Nodes Scenario with Hyper-V cluster" technical paper.
If i have 2 NIC's for sync, 2 NIC's for ISCSI, 1 NIC for heartbeat and 1 NIC for normal network communications and i setup the ISCSI as it lays out in that document, I get no traffic going down the interfaces assigned to ISCSI. This is setup so that the "Active" MPIO path is 127.0.0.1 and the other 2 interfaces are on standby. I could understand this if the 127.0.0.1 path was active on the local machine talking to the local storage and the targets on the second machine you set the active path to one of the actual NIC interfaces but it seems that when you set the "fail over only" policy it changes ALL the targets on that machine to the same interface.
Could you please let me know if this configuration is correct and i am not supposed to see traffic on the ISCSI NIC's? Or do i have the wrong idea when i think that if the paths are set to fail over only i will not see traffic on the NIC's because if a server does fail it would not use any path.

Thanks

DataStream

[thefinkster]

From what I understand about Local-based StarWind with Hyper-V running on top; is that the local iSCSI port is so that StarWind uses the internal iSCSI driver and not touch the network stack, EXCEPT in the case of a failover. I am just getting my Hyper-V Clustered setup (having issues with b7354 sync speeds); but my understanding is that the local is for direct access (super-fast as compared to NIC speeds) and the other nics are for failover (two for redundancy).

On my setup, the failover kept 127.0.0.1 as the primary/active, and the other two NIC paths are failover. I did not see the iSCSI paths change all to 127.0.0.1

StarWind staff; please correct this assertion if incorrect.

[datastream187]

Hi thefinkster, thanks for the reply. I was speaking to a friend today who has experience with StarWind v6 and mentioned to him about the ISCSI setup for v8 (he said in v6 he always setup ISCSI as round robin and showed me a couple of installs he had done, but he had not used v8). He said the way its setup will work fine and the only time i would ever see traffic on the ISCSI NIC's is if one of the host servers StarWind disk/.IMG failed and then that server would use the ISCSI NIC's to talk back to the other servers disk/.IMG. To be honest i didn't actually think of this scenario, i just assumed if the disks/.IMG died the VM's would just be bought up on the other host.
StarWind staff is this correct?

thefinkster sorry i didn't explain it well, yes that's what i meant when i said all the targets change to 127.0.0.1, i didn't mean that all the MPIO paths when you look at the "details" of them actually change to 127.0.0.1 i just meant that if you have 4 targets on your initiator ( 2 local and 2 remote ) and your remote targets have 3 paths under devices -> MPIO ( 1 x 127.0.0.1 and 1 for each NIC assigned to ISCSI ) and you set the active one to 127.0.0.1 then it will mirror this config for all the other targets on that server. So you couldn't have for example your local storage active path as 127.0.0.1 and the remote storage active path as 192.168.0.1, unless i am missing something of course?

Thanks

DataStream

[thefinkster]

Yea, I don't have any device/image pairs pointed to the remote system first; they're all local first then remote via the two NICs. I'll have to try that when I do more testing once I get a new build in my hands.

[Anatoly (staff)]

Hi all!
Yes, you are absolutely correct - traffic that runs through the loopback (127.0.0.1) doesn`t affect the network part at all.
The only way to measure it is to run some benchmark (we usually use IOmeter).