Connection behavior for HA connections

[kmax]

Evaluating Starwind, latest version. HA target on two x64 Win 2008 R2 hosts.

Setup the initiator on another 2008 host. Got MPIO setup. Round robin works, I can verify it by looking at network activity on both the hosts while doing large file copy. They each show half the total activity. That is perfect.

My question is:

When I simulate a node failure by stopping the Starwind service on one of the hosts, the file copy will pause for around 20 seconds. Network activity will go to 0 on the remaining host. After about 15 seconds the file copy continues. The initiator will always show the other host as connected, and of course the other host as "retrying" during this time, which is expected.

Is this normal? Or am I missing some setting somewhere?

Thanks.

[anton (staff)]

This is absolutely normal and it's the way MPIO works. Delay when node is down, node marked as "dead" and pending I/O operation continues.