public_git
/
actions-runner-controller
mirror of https://github.com/actions-runner-controller/actions-runner-controller.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Add docs for setting the RUNNER_GRACEFUL_STOP_TIMEOUT env var on docker container (#2843)

This commit is contained in:
Nathan Heaps 2023-08-29 23:30:18 -04:00 committed by GitHub
parent 5e23c598a8
commit 0b24b0d60b
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
docs/automatically-scaling-runners.md
View File

@ -554,7 +554,7 @@ This can be problematic in two scenarios:
> RunnerDeployment is not affected by the Scenario 1 as RunnerDeployment-managed runners are already tolerable to unlimitedly long in-progress running job while being replaced, as it's graceful termination process is handled outside of the entrypoint and the Kubernetes' pod termination process.
To make it more reliable, please set `spec.template.spec.terminationGracePeriodSeconds` field and the `RUNNER_GRACEFUL_STOP_TIMEOUT` environment variable appropriately.
To make it more reliable, please set `spec.template.spec.terminationGracePeriodSeconds` field and the `RUNNER_GRACEFUL_STOP_TIMEOUT` environment variable appropriately. **NOTE:** if you are using the default configuration of running DinD as a sidecar, you'll need to set this environment variable in both `spec.template.spec.env` as well as `spec.template.spec.dockerEnv` for RunnerDeployment objects, otherwise the `docker` container will recieve the same termination signal and exit while the remainder of the build runs.
If you want the pod to terminate in approximately 110 seconds at the latest since the termination request, try `terminationGracePeriodSeconds` of `110` and `RUNNER_GRACEFUL_STOP_TIMEOUT` of like `90`.