Kubernetes Deployment Tuning

3 posts · May 2023 – Mar 2026

What this trail covers

EvenUp migrated from a self-managed workflow execution system to Temporal, leveraging Temporal's built-in metrics for scaling workers based on task slots and latency. They implemented a Kubernetes liveness probe using Temporal SDK metrics to automatically restart unhealthy worker containers, improving reliability and reducing on-call incidents. This contributes to a more robust and observable worker deployment strategy.

All Temporal trails

Mar 2026When running your own orchestrator isn’t worth it anymore — and why EvenUp chose TemporalEvenUp details their migration to Temporal, focusing on their use of `temporal_worker_task_slots_used` and latency-based metrics (`temporal_workflow_task_schedule_to_start_latency`, `temporal_activity_schedule_to_start_latency`) for scaling workers. They also describe implementing a Kubernetes liveness probe using metrics like `temporal_num_pollers`, `temporal_worker_task_slots_available`, and `temporal_request{operation="RespondWorkflowTaskCompleted"}` to automatically restart unhealthy worker containers, eliminating periodic processing slowdowns and on-call pages.
Dec 2025Tips for running Temporal on KubernetesThis post provides practical tips for deploying and operating Temporal on Kubernetes to achieve predictable latency, safe upgrades, and efficient scale. It covers running the Temporal Service using Helm charts, deploying Temporal Workers, and autoscaling Workers based on Task Queue backlog and Schedule-to-Start latency. It also discusses planning capacity and tuning for production, including addressing CPU throttling and the importance of load testing.
Jun 2023Kubernetes CPU ThrottlingThis post provides actionable advice for tuning Temporal Server deployments on Kubernetes to reduce request latency and improve node efficiency. It explains Kubernetes CPU throttling, the impact of `GOMAXPROCS` on Go applications, and recommends disabling CPU limits in favor of CPU requests. It also offers strategies for mitigating latency spikes during Temporal upgrades by adjusting Kubernetes rollout parameters based on the number of history pods.
May 2023Scaling Temporal: The basicsThis post details a load testing process involving adjusting load, measuring performance via state transitions and request latency, and scaling Kubernetes resources and Temporal shard count. It highlights the importance of persistence backend as a bottleneck and the impact of shard count on lock contention and memory usage. It also discusses the trade-offs of increasing shard count, such as increased memory usage in History pods, and the current limitation of not being able to change shard count after cluster creation.
Advertisement