VMX 6.1.0.52798 (Alpha 1)

IMPORTANT NOTICE FOR ALL USERS:

By using this software you accept responsibility for all charges that are incurred with your cloud provider(s).

Please read the following thread for important information regarding VMX:
viewtopic.php?f=189&t=10267

This Alpha of the VMX extension for Deadline is compatible with Deadline 6.1.52776 (Beta 5). Please read the following threads for important information regarding Deadline 6.1:
viewtopic.php?f=85&t=4484
viewtopic.php?f=85&t=4483

Installation

  1. First make sure you have installed Deadline 6.1.52776 (Beta 5).
  2. Download the .zip file attached below and extract it.

The .zip file contains the Alpha build folder that has all the files necessary to run the VMX Balancer. The recommended location for the Alpha build folder is in the “C:\Program Files\Thinkbox” folder. For example, right-click the .zip file and choose “Extract All…” and then browse to the “C:\Program Files\Thinkbox” folder. DO NOT extract these files on top of your Deadline6 folder. Inside the Thinkbox folder you should see the Deadline6 folder (where Deadline 6.1 Beta 5 was already installed) and also the VMX Alpha 6.1.0.##### (Alpha #) folder. The Balancer can be run by double-clicking the deadlinebalancer.exe file inside the bin folder.

Note: Deadline 6 Beta 5 includes only the Amazon cloud plugin. If you would like plugins for other providers, please post a request or e-mail me directly. Also, the Asset Crawler script is not included in the above zip file. I will post it separately.

Attached is a copy of the server-side asset checker script we wrote for VMX.

This script basically just listens on a port (default is 53544, you can change this by editing the script) for requests from the Balancer. The Balancer will send it asset paths, and the script checks if they are accessible from its current location (returning true or false). The idea is to determine whether or not required Assets are available on the Cloud before spinning up VMs for a Job, to just proceed to fail on missing assets over and over.

There are a couple things to note:
For this to be useful at all, you need to be specifying required assets for Jobs at submission time.
You also need to make sure whatever machine you run this script on will be seeing files the same way Cloud Slaves in that region will be.
The Balancer obviously also needs to be able to talk to the machine running this script; that able to resolve its hostname/IP, and allowing the port through firewalls (again, 53544 by default). These settings are both configured in the Balancer tab of the Cloud Region setup.
Currently, I believe this will only affect the actual weight of Groups when determining which to spin up; Slaves might still be picking up Jobs with missing assets. We should be adding a check to the Slave for this when dequeuing tasks, soonish.

It’s just a python script, so serverside you just need to run “python </path/to/script.py>”, assuming Python has already been set up.

Cheers,