How to deploy multiple instances sequentially?

[darrenoid]

Hello,

I am trying to find a way to deploy each instance on a multiple instance deployment sequentially. What I have noticed is when I choose more than one instance on a request, each instance is processed at the same time. So, if I choose 12 instances and the blueprint is using a clone workflow, then 12 clones will fire off at once from the same template. This can hammer the datastore. I would like to find away to make each instance deploy one at a time. I have not been able to find a custom property for this.

So far my workaround has been to drag multiple machines onto the blueprint canvas and then make them dependent on each other. This forces the machines to deploy one at a time. However, dragging 12 machines onto a blueprint and then editing the custom properties on each one is very tedious and time consuming. There is no way to copy and paste a machine with all it's properties on the canvas. It would be much easier to have one machine on the blueprint, but deploy multiple instances.

Any help on this would be appreciated.

Thanks,
Darren

[daphnissov]

Just to be clear, is your sole reason for wanting to do this because of the high datastore activity that concurrent clones generate?

[darrenoid]

Hello daphnissov,

Yes, that is the main reason. It also makes the monitoring logs easier to follow for me without having to filter for individual machine names.

Regards,

Darren

[Jiju_25]

By default vCenter supports 8 concurrent operations. vRA by default supports 2 concurrent provisioning. If you want to change the default provisioning from 2 to someother value after taking into consideration the endpoint[vcenter in this case] can handle it then you can change it accordingly. More details over here

Increasing vRA's Concurrent Provisioning Operations - [virtualjad.com]

[darrenoid · Accepted solution]

I found a decent workaround to accomplish this. I fired off a deployment with 1 machine and then afterwards I used the scale out feature to add more machines 2 at a time. This worked well for me. It was also nice when a scale out request failed, I was able to just fire off another one while keeping the main deployment in tact. If I had deployed all instances at once in a single request and one of them failed, the whole deployment would have been destroyed.