Load Test

[daniel.witter]

Afternoon,

We are moving to the high availability model but we just wanted to know if it’s possible to run load tests on the Octopus server so we can predict when we would possibly need to add another node further down the line

I understand this could be tricky to answer as every project is different varying in sizes and processes.

Regards,

Dan

[mark.reeder]

Hi Dan,

Thanks for reaching out! That’s a good question, and I agree it is a tricky one to answer as there are many variables. In theory, load testing is possible as Octopus is created API first, and there are plenty of libraries out there to perform UI actions as well. However, I can’t say I’ve heard of any other customers doing load testing to try and make that determination.

There are a few different levers you can pull which will affect the performance of Octopus, especially with a HA setup. For example, if you are noticing UI slowness you could convert one of your nodes to a UI-only node by setting the task cap to 0. If you start to notice tasks queueing more than is desirable and the hardware Octopus is hosted on can support it, you can increase the task cap. In that regard, yet another lever may just be to increase the VM size/resources Octopus is hosted on.

We have a write-up in our Best Practices guide which goes through some different configurations for varying sizes/use-cases and that could be a good start. You may want to consider the type of teams that will be using Octopus and the deployments you will experience on average. Also, take into account any external scripts which may be hitting the API directly, if any. With that info, you could probably set up a load test using whichever method you find to most accurately portray your situation. As this may mean you are creating releases and performing deployments, it might entail creating another environment to test with to accurately depict how tasks occur. As you may see, the scope of this could grow quite rapidly. I’d recommend using the experience you have with your non-HA set up along with the Best Practices guide to help guide you as you monitor and adjust over time.

I hope this helps!

Thanks,
Mark