Hello Octopus Support,
I’m using a worker pool for my deployments and I have noticed the task distribution is not being evenly assigned to my linux tentacle worker nodes. Looks like one out of the two nodes in the pool is getting 80% of the tasks. I was looking to find some guidance with how the task/load distribution should function on the worker pools or perhaps some configuration I missed? I’m running Octopus Deploy version 2023.2.12868
Hi Mario,
Thanks for getting in touch! Octopus doesn’t currently balance worker task distribution based on load. We have some documentation on our workers which goes into more detail on how we handle worker leases. Our Worker pools docs has a small section on this but we go into further detail in our worker configuration best practice guide, Offload Work to Workers.
When you do a deployment or a runbook run with workers, a worker is leased from the pool; the work is done, then the worker is added back into the pool. The vast majority of the time, the same worker is used for a single runbook run or deployment. But the worker can change in the middle of the deployment; you should design your process around that assumption.
Essentially, there’s no load/task balancing going on, Octopus will grab the first available worker and use it, regardless of other workers in the pool.
There has been some talk internally about whether we will introduce some form of load balancing to worker pools, but it’s not something we’re currently working on.
Let me know if that helps or if you have any further questions or thoughts on this.
Best regards,
Daniel