failover ip

[aclhkaclhk]

does DSS support active/passive servers with failover IP so that iscsi clients connection remain unbroken during failover?

[techtou]

does DSS support active/passive servers with failover IP so that iscsi clients connection remain unbroken during failover?

Same question here.
When performing failover all our virtual servers end up restarting.
Hosts running them report that iscsi connection and storage was dropped and regained within a second - enough for all virtual machines running from the SANs crash and reboot.
Is there a way to tune this so the failover is seamless and storage is available 100% of the time during failover/failback?

Thanks.

[techtou]

Anyone care to address?

[To-M]

We are working on having an Active / Active solution this later part of this year.

[AdStar]

Same question here.
When performing failover all our virtual servers end up restarting.
Hosts running them report that iscsi connection and storage was dropped and regained within a second - enough for all virtual machines running from the SANs crash and reboot.
Is there a way to tune this so the failover is seamless and storage is available 100% of the time during failover/failback?

Thanks.

Hmm I don't get this when I failover, ESXi 4.1 MPIO (4 nics), I can failover/back without any of the VM's rebooting or LUN timeouts etc. It works very well. I might get 1 dropped PATH on ESXi but by the time it poll's and recovers failover is completed, and everything hum's along as it was.

What is your configuration?

[Gr-R]

I was wondering the same thing AdStar.

By design, failover is near instantly. Dropping hosts on every failover is certainly odd.
Some things do come to mind though;
Is everything the same with regards to the targets on each node: frame settings, thread queues, etc. What about frame sizes on the NICs?

And as To-m mentioned, active/active is coming, so another failover option will be available.
But dont give up on node to node failover.... seems something just isnt right in your configuration someplace.

[SeanLeyne]

We are working on having an Active / Active solution this later part of this year.

Todd,

It is later in the year... any news on this?

I am having to investigate other software solutions (Nexenta) in search for an active/active failover.

The current Open-E solution is too problematic at the moment.

[To-M]

What version are you using w/ V6 last 2 build we have had really good results due to the new IO error detection that we have.

Active / Active had to have a whole new architecture so it will be a newer product due to the advanced features so it will be in Q1 of this yr - beta to be out next month then we need 2 or 3 months to test.

[stevens_pete]

Todd,

We are looking at Active/Active as aopposed to Active/Passive do you have any ideas to when this will be coming up!

Cheers

Pete

[To-M]

We hope to have all the tests completed in Q1 of next year and then released in Q1 but cant give an exact date due to the amount of tests that will need to done.