Upgrading NAS - How can i increase iSCSI Keep alives or time outs-while it reboots internally

[BIACS201110141]

I am running an EMC Celerra NAS.  VM's are running over iSCSI.  While the storage is upgraded the front end will failover on the NAS, so the disk is not accessible for about 4-5 minutes.

Where do I turn up the ISCSI values to allow for an extended pause durring the upgrades?

[DSTAVERT]

Welcome to the Communities.

An active virtual machine would normally be regularly writing to disk. I assume this is a single controller Cellera. You might be able to get a powershell script to run prior to your update and pause the virtual machines and then un pause them afterwards.  This would cause the virtual machines to do something similar to hibernate on a laptop where active RAM is written to disk. It may not be the best approach for some applications. Otherwise I think you are out of luck. I would check with your EMC support folks for any suggestions.

[fgl]

If this is a single controller Celerra NAS then your best options are to either temporary suspend all your VMs or schedule a window of downtime to shut everything down which I feel is safer.

If this is a dual controller Celerra NAS then you should be fine as I have done Celerra upgrades, reboots, failovers, and failbacks with about 100 VMs running off it via iSCSI and none of the VMs ever went down or had any issues afterward.

[BIACS201110141]

@fgl  do you mean dual SP's?  otherwise not 100% sure where I would locate that info.

[DSTAVERT]

If you have dual storage processors, one is upgraded and the device fails over to the second processor. You should have no downtiime. If it is a single then you would have the downtime.

[fgl · Accepted solution]

Yes. I meant dual SP's.  Sorry for the confusion as I've worked on EMC, NetApp, Compellent, etc... and different vendor use different terminology for the same thing.

[BIACS201110141]

When the front end data movers failover there will be downtime -But im looking for real world experience.

[fgl]

The same should still apply unless you have only a single.

From my real-world experience I have gone through several Celerra DART upgrades as well as Clariion FLARE upgrades and not have had any issues where it brought the VMs down but then again we schedule these upgrades during hours of minimal usage.  If you are still concern and want to be on the safe side then I suggest you see if you can schedule a complete shutdown of all going to be affect VMs.

I just thought of something you didn't mention if you are using software iSCSI or hardware iSCSI HBA so I just assumed you are using software iSCSI since you are talking from the Celerra end. Software iSCSI is what I am using and other places I've worked at also used software iSCSI unless they where doing FC to the SAN which it didn't sound like you are doing.  But in any case if you do have HBA's connected then you will most likely lose connection and I don't know any easy way around that.

I'll tell you about another option I did but this has nothing to do with SAN or NAS code upgrades but with a scheduled complete power maintenance in the data center in which all power was going to be down for 12 hours and it might give you some ideas.  Lucky we had a few ESX hosts that has a lot of local storage enough to svmotion all of our production VMs to and I moved those ESX hosts to another building before hand where the power was not going to affected and everything production wise stayed up during the 12 hours outage and afterward I svmotion everything back and everyone was happy including the top execs.  I have all of our production VMs on blades across 2 chassis but I have a few standalone hosts for redundancy in case and for standalone testing which is why these hosts has a lot of local storage, which in this case they saved me and made me looked good to the brasses for creative thinking now I only wished they gave me a long needed raise instead of a 'keep up the good work' spiel.