We have a cron job that cleans up things that were missed and we missed a lot in this repo https://github.com/heroku/hatchet/actions/runs/4409068091/jobs/7724891614:
Destroying apps older than 10m
Destroying "hatchet-t-2b560971d6": d00b46ec-f786-4bf3-a80c-8eac60fb6ab6, (11/20) reason: app age (288.26m) is older than 10m
Destroying "hatchet-t-943260842d": 0b7b8093-b0ad-4eae-bb1f-8643ffd20388, (10/20) reason: app age (288.27m) is older than 10m
Destroying "hatchet-t-9ad2d761fd": 831b05a0-0a99-492f-95bc-3cf287091f65, (9/20) reason: app age (288.27m) is older than 10m
Destroying "hatchet-t-7171038be6": 35c9faba-da01-443a-b1ab-6cd64719694b, (8/20) reason: app age (288.23m) is older than 10m
Destroying "hatchet-t-e226788c87": 4d932982-d756-4105-b125-6002cf21b76c, (7/20) reason: app age (288.23m) is older than 10m
Destroying "hatchet-t-fea0f82c3f": a5243ef0-387a-416e-8e9e-9ca83fc5e207, (6/20) reason: app age (288.23m) is older than 10m
Destroying "hatchet-t-033d285733": e7c86a90-95d5-4fd0-a36f-1fcc4943ffdc, (5/20) reason: app age (288.2m) is older than 10m
Destroying "hatchet-t-253e02155d": 4cae07f9-7b38-4041-be86-7d2d0273773c, (4/20) reason: app age (288.21m) is older than 10m
Destroying "hatchet-t-76264f5df0": d8cbc1be-ff25-47fd-8b8d-dde78e98fe51, (3/20) reason: app age (288.21m) is older than 10m
Destroying "hatchet-t-78b524b587": 289f1441-bc4f-4e4e-947c-26f5cf3b01f2, (2/20) reason: app age (288.22m) is older than 10m
Destroying "hatchet-t-8085b919b6": 832e981c-0b4d-4378-b783-603207a0af2e, (1/20) reason: app age (288.22m) is older than 10m
Destroying "hatchet-t-ea38fa3ff6": 56fe4db9-7322-4a62-a8cd-a57744ed0561, (0/20) reason: app age (288.22m) is older than 10m
Done
I do some mocking/stubbing in this library that might have unintended side effects. If we wanted to chase it down I would: Add logic to print a backtrace and app name for each app being created. Then after the tests run, see what apps are left alive and you can use the backtrace from the logs to figure out what tests are causing the leaks
It's not an emergency and can wait. I'm leaving this note to a future coder (who might be me).
We have a cron job that cleans up things that were missed and we missed a lot in this repo https://github.com/heroku/hatchet/actions/runs/4409068091/jobs/7724891614:
I do some mocking/stubbing in this library that might have unintended side effects. If we wanted to chase it down I would: Add logic to print a backtrace and app name for each app being created. Then after the tests run, see what apps are left alive and you can use the backtrace from the logs to figure out what tests are causing the leaks
It's not an emergency and can wait. I'm leaving this note to a future coder (who might be me).