@Copyright: I had the same problems on my free vSphere 4 ESXi at home you have when I tried to install the ASG's on "Other 64-bit Linux" with E1000 NIC's. But as I already said it was fine after using "Other 32-bit Linux" with "Flexible" NIC's.
@da_merlin: I was using an independent and dedicated virtual switch to connect the HA interfaces. I'll re-produce the failed setup later on today and post the master's HA-log.
@da_merlin: Attached the high-availability.log from a fresh and failed cluster installation using the 7.902 iso image on vSphere 4 ESXi (other-Linux-64 with E1000 NIC's connected to a dedicated vSwitch). As Copyright said the slave shutdown after a few minutes...
This is an vmware issue, the e1000 implementation of vmware is unable to handle jumbo frames.
If HA/Cluster detects Gigabit Ethernet, it will set the MTU size to 2000. During startup of a HA/Cluster Slave the MTU will be validated. If this fails, the node will automatically shutdown.
@Copyright: Whats in the high-availability.log on the master during slave startup?
Is there another HA/Cluster system running on the same VMWare server? Each HA/Cluster system needs an dedicated virtual Switch for the sync interface.
da_merlin, I am running 2 x ASG320 HA productive (physical cross-ethernet link for sync) and 2 x ASL 7.902 VMs on top of VMware vSphere 4 (dedicated VLAN link for sync) for testing.
But all 4 instances use the same LAN(subnet), and so I see HA sync messages "version conflict" if I start HA/Cluster on the ASL 7.902. This is ok, because all 4 instances use the same LAN (subnet).
Is an seperate virtual Switch needed for clean HA sync on top of VMware vSphere or ist an VLAN also ok ?
