PCNS Configuration Question & Issues

[cresch]

I deployed the PCNS appliance (new v5 licensed version), added my (2) hosts supported by the UPS, etc. and all looked good.

The shutdown parameters were showing as follows:

Low battery duration currently set to 20 mins
Shutdown delay FOR NOW set for 240 secs (4 mins). 
Maximum required delay currently set to 14 mins (based on VM shutdown times selected, I believe)

I then performed a shutdown test by cutting the power to the UPS.  Unit was showing 30 min runtime at start of testing.

Approximately 10 mins later, UPS was reporting low battery with 20 mins remaining (which would match with configuration I set).

Although I can't find anywhere to see the log or see what's being triggered (maybe there are PCNS logs somewhere I could see?), I noticed the VMs were instructed to shutdown on the hosts, and did so...with the exception of the PCNS VM (on one host) and the VCSA (on the other host).  

At some point after, host 1 with the PCNS appliance shutdown completely.  However, host 2 with the VCSA not only remaining running, but the VCSA was still running as well (so it wasn't shutdown).

For testing purposes for full power-on from off, I went ahead and shutdown the vCenter and then the host manually, and then turned off the UPS load manually for force complete power off for all attached equipment.

After a minute or two, I turned the UPS load back on.  SAN equipment powered on as expected, following by hosts after 5 mins (since PDU includes a 5-min delay to allow SAN/storage to come online before hosts attempt to access).  However, at some point, both hosts then powered themselves off, and remaining that way until I manually powered them back on.

As a result of the testing, I have several questions and issues about the actual & expected operation using PCNS in the environment:

1. Do I need a PCNS appliance deployed to each host that that the vcenter is local to the PCNS appliance and it will properly shutdown the VCSA first before itself? (or maybe some misconfiguration that caused the behavior I saw)

2. I'm guessing that the hosts powered OFF after the restart possibly due to the PCNS maybe coming online and then completing pending tasks or something?  (otherwise, no idea why the hosts would have shutdown again after initial boot). Again, any log files I can review for the PCNS-triggered actions would be helpful.

3. After manual power on of the hosts, no VMs automatically powered on.  This is somewhat expected in an since the built-in power-on settings aren't officially supported in an HA environment.   However, since this obviously then requires manual intervention, I'm wonder if anyone has found a way to use startup scripts or such to force the VCSA and/or other key VMs to start up when an HA environment is powered on from hard down.

As compared to a hard crash prior to having PCNS, most of the VMs were shutdown safely, but now it's a much more involved startup process to get the environment operation again (vs when they hard-crashed, the HA components kicked in when the hosts restarted and they automatically attempted to restart most guest VMs), so just wondering what I'm missing to help make this a complete solution.

If anyone can provide insight as to expected operation, configuration issues, log files, or start-up scripting related to the above observations, I would greatly appreciate it.

[BillP]

@cresch 

I recommend you watch our Youtube video. How to install and configure PowerChute Network Shutdown for VMware.

https://www.youtube.com/watch?v=JzsDuV3oMxc

Questions:

1- No

2 - The shutdown process had not completed so when PowerChute was restarted it completed the tasks. The reason the host did not power off was because there was a VM still running. 

3 - PowerChute can be configured to restart all VMs it powers off. See the video. 

All logs are store in /opt/APC/PowerChute/group1. See Schneider Electric FAQ FA241061

Or read the User's Guide.