we have opend a Ticket already with EMC , but as we cant use our shiny new Unity at all at the Moment . I would like to adress the Problem to the Community as well.
Ist Situation , we have a VNX connected to a Hyper-V 2012 Cluster via ISCSI ( 10GB ) ... MS DSM was used not PowerPath . The Combination workst since 4 Years ... no Drops not Problems ... rock solid !!ur
Now we got our new Unity the first Thing we noticed that the Lun 0 is now not a Disk -1 as on our 2 VNX .
We configured the ISCSI Part exactly ! like we did with the VNX , everthing looks alike .
BUT if you switch the Hyper-V Disk Role to another Node in the Cluster the DiskRole Process will take Seconds until the ROLE is GREEN again in the FailOverCluster Managment Console and even then there is now IO Traffic until around 80! Seconds later. "
So the VM "Stops" for 80 Seconds ( like it is freezed ! ) . The Storage is for that Reason at the Moment not usable because we have such an Impact in the Production that it is irresponsible to use the "expensive" Unity for that Matter.
The Same Disk Role Change on a Disk Role with a VNX 5600 then the Traffic will "break in" for a fraction of a second and then it "goes" on ... like a breeze.
Anyone has such an Experience also in Regard to an UNITY 550F, 300 ( firmware 4.4 ) . The two Unitys are connected via Each other in an active/active Failover Senario ( for that Reason then need direct FC Connection ). We have a Speciality on the UNITY 550F as they have the ISCSI ( Ethernet Interfaces with Offloading ) . Maybe there are also Question in Regard to PowerPath and the "ISCSI with Offloading" , which could fit the Problem Description.
Maybe someon has a Glue where to start the Search or has maybe heard form EMC ( I read about PowerPath Error Description which sounded alike )
Regards and Thanks
Did you follow the instructions in the Configuring Hosts to access FC or ISCSI storage PDF manual and looked at the Unity Hyper-V white paper ?
Of course we did.
We did a lot of the testing now on our own , as we stick for some reason in an Support / Engeneering "Support" Problem.
We found out so far, that if you connect the same Cluster to a UNITY Hybrid ( without Offloading ) the Problem didnt occur.
When the Diskrole is switched then we have the same behaviour as we see in our two VNX .
So back to testing on the Unity 550F, we checked every SFP , every Cable and every switch it is attached too. ( No Result ! )
We unplugged the SFP Devices to the SP and finally we found some weard behaviour. If the Storage is only connect via 3 Path , then the Problem doesnt show . Hey Great !!
We asks EMC for a Replacement of the Offloading Module , we received the Replacement , but strandgly the Problem persist on the same Port which is tight to a SP, so the Problem might sits "higher" -> points to the SP ....
Well lets see what Engeneering is doing with those new Infos now , Technik is interesting and so unexpected
Now Engeneering changed also the ServiceProcessor, the Problem is now on another Port . So the understanding is now , that it is a Problem in the Firmware.
So now we can state that if we unplug that "particular" Port , the Unity 550F will not show the Problem . Its interesting to tell that those "Effects" are not seen on a UnityVSA or an Unity Hyprid 300 .
which firmware - do you have different Unity OE on the AF than on the hybrid ?
SR number ?