Someone copied their local command into the pipeline, including --headed, and the job now hangs and is killed at the 60-minute timeout. Removing the flag fixes it, so the immediate problem is solved, but I would like to understand what it was waiting for — and whether there is a legitimate reason to run headed on an agent.
Accepted answer
Headless is the default and it is the right default for an agent. --headed asks for visible browser windows and a build agent has no display, so the browser cannot launch at all.
There is no legitimate reason to run headed in CI, because the thing people want it for — seeing what happened — is served better after the fact:
.sdods/runs/<runId>/, uploaded as artefacts,Keep --headed, --ui and --debug for your own machine. They are all interactive.
Have a different question? Ask it.