I’m having an interesting problem with some of my builds, such as:
I run the same tests on Drupal and WordPress. For some reason, the WordPress builds are always failing; however, I can’t tell why, because Circle always says that the problem is with the container, and it restarts the build. The next time it runs, it fails, because it tries to create a site named after the build number, and Circle re-uses the build number. The larger problem is that I don’t get to see any of the output from the first build.
There shouldn’t be anything different between the Drupal and WordPress versions of these tests, so I’d almost write this off as just being a random Circle hapenstance that I should code around and ignore. But it only seems to happen on WordPress, and WordPress hasn’t passed for a while.
It would be easy for me to fix the naming problem if there was a sequence number I could use to tell if the build had been restarted. I guess I could reduce CIRCLE_WORKFLOW_JOB_ID a bit and use that, or maybe just use the current time. I sort of wonder if I am going to cause an infinite loop if I fix the symptom.
Any insights appreciated. Closing as “behaving within operating parameters” is also okay.