Improve the documentation and error reporting related to node_tolerations
Description
If one adds the following config to config.toml
for a runner with kubernetes executor according to the keywords section,
[runners.kubernetes.node_tolerations]
"dedicated" = "gitlab:NoSchedule"
CI jobs could fail with the confusing error message:
ERROR: Job failed (system failure): Pod "runner-9d53a96d-project-434-concurrent-08g66g" is invalid:
spec.tolerations[0].effect: Unsupported value: "NoExecute": supported values:
"NoSchedule", "PreferNoSchedule", "NoExecute"
Proposal
- We need to provide better examples in the keywords section about how to set the value of
node_tolerations
- It would be also good to improve the error message and make it more straightforward
Links to related issues and merge requests / references
Ticket from a customer: https://gitlab.zendesk.com/agent/tickets/118399 (internal only)