Clean Shutdown of Starwind Cluster

Software-based VM-centric and flash-friendly VM storage + free version

Moderators: anton (staff), art (staff), Max (staff), Anatoly (staff)

Post Reply
BenM
Posts: 35
Joined: Wed Oct 24, 2018 7:17 am

Thu Jan 30, 2020 12:13 pm

Guys,

Unbelievably, the power company has given us notice of power interruption.

I know we should have UPS to cover this but we don't, so I will be shutting down the cluster & the hosts in advance and restarting after the power interruption.

Am I right in thinking that if I stop the cluster using the Fail Over Cluster Manager (or powershell stop-cluster), wait a while for any writes to clear through then shut down the nodes one at a time, Starwind will know what to do when the system is brought back up?

I guess as long as I power them back on in reverse order to shutdown everything will pop up just sweet. - for example if I shut down node 2 then node 1 I would restart 1 then 2 (I can enforce this by yanking the power leads from 2 when it is shut down)

Ben
yaroslav (staff)
Staff
Posts: 2277
Joined: Mon Nov 18, 2019 11:11 am

Thu Jan 30, 2020 1:11 pm

Greetings,

Here's how you can avoid the full sync.
1. You need to shut down gracefully all the virtual machines.
2. Set all the StarWind devices to the Maintenance Mode. Here is the article on how to enable Maintenance Mode: https://www.starwindsoftware.com/help/M ... eMode.html.
One more time, make sure to shut down all the VMs before setting devices to Maintenance Mode.
After StarWind devices are under maintenance, feel free to shut down the servers.
BenM
Posts: 35
Joined: Wed Oct 24, 2018 7:17 am

Thu Jan 30, 2020 2:42 pm

Yaroslav

Great thanks for that....

So now the sequence is:-

Gracefully shut down all guests
stop the cluster
Put Starwind in Maintenance Mode - using example starwindX\powershell\MaintenanceMode.ps1 as a starting point.
Shut down cluster nodes
power off
<wait for power company to do their thing>
Power on
reboot cluster nodes
SW out of maintenance mode
start cluster
Boot the guests

Thanks for your help - probably saved a bit of stress the other side of the power outage and certainly some before it as well.

Ben
yaroslav (staff)
Staff
Posts: 2277
Joined: Mon Nov 18, 2019 11:11 am

Thu Jan 30, 2020 3:59 pm

Ben,

You do not need to reboot the nodes, just start them as usual.
The sequence you outlined looks great.
BenM
Posts: 35
Joined: Wed Oct 24, 2018 7:17 am

Mon Feb 17, 2020 2:16 pm

To wrap up the thread...

I noticed a couple of gotchas....

1. using powershell/StarwindX to put the HA images in maintenance mode - message received that there were still clients connected to the devices, which, I guess, there were in that the stopped cluster iSCSI had connections to the drive. So just used Force=$True to set maintenance mode; after all, there was no process on the servers that could write to the SW volume (other than SW). Unsetting maintenance mode via powershell did not require the use of Force=$True.

2. On reboot the cluster service autostarted and propmptly autostarted the cluster - not a biggy becasue with SW in Maintenance mode connections to the CSV were denied.... cluster consequently moaned about lack of CSV connection until I could get the drives out of Maint mode. Should have set the Cluster service to manual rather than leaving it on automatic. Hmmm That almost sounds like a good idea.

ttfn
yaroslav (staff)
Staff
Posts: 2277
Joined: Mon Nov 18, 2019 11:11 am

Tue Feb 18, 2020 1:07 pm

Greetings Ben,

Thanks for sharing your experience with others. Should you need any assistance in the future, we are here to help you.
Post Reply