I have a single hypervisor - a HP ProLiant DL380 G7 with two six-core Intel Xeon X5690 CPUs - running ESXi 6.5.0 5969303. I am running vCenter Server 6.5 in a Windows VM on that hypervisor.
As part of my evaluation of VMware I run a number of Linux 'test' VMs - all cloned from the same Red Hat Enterprise Linux 7.2 VM - which execute a number of tasks in parallel. With one particular task one of the Linux VMs will usually hang. The task is a build of a Linux kernel, something that takes around 13 minutes on a single Linux VM.
For example, in one test run I had four of the Linux VMs all running four compilations each in parallel, and one of the VMs hung. In another run I had six Linux VMs each running two parallel 'make linux' commands and one of those VMs hung. In any scenario where I'm running a number of VMs and the VMs are executing one or more kernel compilations one of the VMs will usually hang.
By 'hang' I mean -
* The Linux VM is totally nonresponsive - won't ping, VMware remote console shows the login prompt but won't respond to keys/returns, seems dead.
* vSphere Web Client - running from the Windows vCenter Server VM - doesn't show any problems with the hung VM - it says it is 'powered on', all green.
* The 'Magic SysRq' key works for the Linux VM from the console - I can initiate a crash dump with Alt+SysRq+c.
I have only one thought as to the source of this problem - the hardware, a HP ProLiant DL380 G7, is listed in the VMware Compatibility Guide as supporting only up to ESXi 6.0 U3. I'm used to applications being more 'forgiving' of slightly older hardware ... but is this a legitimate and realistic excuse/reason for my 'hanging' problem?
The Proliant DL380 G7 was the only machine we could spare for our evaluation, even though we have newer machines for official deployment of our chosen virtualisation platform. As it stands I can't recommend VMware when VMs regularly hang. Since VMware won't offer support without a paid licence I would appreciate any advice from this community on how I should view these VM hangs. Is this sort of thing relatively common, to be expected, when moving beyond the boundaries of the VMware Compatibility Guide? When the difference is all within a 'major' version of ESXi (6.0 U3 vs 6.5)?