Hello again,
I talked with the support rep again and I believe she's touched base with you as well. For the record, my SR # is 12239239310. I wanted to add some more details to this post on things I've learned so far:
1) Our ESXi 4.1 U1 hosts seem to have this value set to 60 secs. My guess is that 60 secs is the default value for ESXi 4.1 U1 and you've sort of confirmed that now.
2) A fresh build of ESXi 5.0 U1 has this value set to 14 secs. My guess is that 14 secs is the default value for ESXi 5.0 U1.
3) Putting a brand new ESXi 5.0 U1 host into an HA/DRS cluster does not change this value, at least not immediately.
4) Creating a host profile from an ESXi 5.0 U1 host that has this value set to 14 secs results in a host profile with no entry for "Misc.HeartbeatPanicTimeout".
5) Creating a host profile from an ESXi 5.0 U1 host that has this value set to 60 secs (or likely any value not equal to 14 based on my testing) results in a host profile that DOES contain an entry for "Misc.HeartbeatPanicTimeout".
6) Some of our clusters contain hosts that are out of sync because of our vSphere 5 upgrade process. That process is:
a) Perform a fresh install of ESXi 5.0 U1 on the first host -- reasoning is to get all the new "defaults" for ESXi, such as this setting, vSwitch port count, etc.
b) Create a brand new host profile from this new reference host.
c) Use VUM to upgrade subsequent hosts in the cluster.
d) Apply host profile to newly upgraded host.
So, this process explains how some of our clusters are mixed up. The first host has the default value of 14 secs, but since that's the default (apparently), vSphere doesn't both to create an entry in the host profile for that advanced setting. Then we use VUM to perform an in-place upgrade on the next host, which is ESXi 4.1 U1 and thus has a heartbeat value of 60 secs, and since it was an in-place upgrade, it retains that value. Applying the host profile does nothing to change that as there is no entry to do so for the "Misc.HeartbeatPanicTimeout" setting. I verified this as we upgraded a simple two-host cluster today that was still running 4.1 U1.
This leaves me with three more questions, which I hope to gather from this SR / community post:
1) What exactly does the "Misc.HeartbeatPanicTimeout" advanced setting do?
2) Why doesn't vSphere create an entry in the host profile for this setting all the time?
3) Since host profiles doesn't always create an entry for the default value, why doesn't an in-place upgrade change the default to the new value (60 to 14)? I mean, if I had changed that value to something other than the old default of 60 secs, then I'd expect an in-place upgrade to respect that value. But if I left it at the default setting, one would assume that I'd want to have the new default for ESXi 5.
