We have a 4 node ASA C800 MCC-IP where everything is connected using 100G Cisco equipment. (please ignore that this is a MC)
We have a VMWare cluster connected using iSCSI. We see 4 paths for each datastore we have created, which is as expected.
If we create a VM on a connected datastore that is located on one of the nodes, and run a bit of load, then do a failover, we see two of the paths go down inside vmware (as expected), and the load continues without any issues... When we do the giveback however, we see a timeout af 40 secs where the VM seems to stall... then continues after that....
Any idea why this is? Surely 40 secs is too much?
I just want to point out that this is an ASA, so we have 4 x Active I/O paths unlike a AFF/FAS where the local paths to the volume would be Active I/O and the paths from the other node would be Active (Passive).
#Cluster HA Giveback results in a 40 sec. timeout on iSCSI hosts
final skiff
ashen notch
The target for ASA is under 30s, but as always is load dependent. Were the systems busy at the time?
final skiff
There is only one VM running iometer with a load while we do the switchover (which is OK) but giveback gives this "timeout" of 40 secs.