Cronjobs failing due to `flock` not obvious

[rocketeerbkw]

In #192 we prefixed all cron jobs with flock to prevent concurrent runs. When flock errors there are no logs so it's not clear if that or the cronjob itself is failing:

time="2025-08-28T10:31:00Z" level=debug msg=executing command="flock -n /tmp/cron.lock.747a6372b42e78c58e130f4618aad7293d6d92096010ff5df770ee08cd5712bd -c 'drush elysia-cron run @default'" spec="* * * * " user=root
time="2025-08-28T10:31:00Z" level=error command="flock -n /tmp/cron.lock.747a6372b42e78c58e130f4618aad7293d6d92096010ff5df770ee08cd5712bd -c 'drush elysia-cron run @default'" result=error spec=" * * * *" user=root

The only way to diagnose this is to see that the time between logs is very short (~6 ns) and to run ps aux to see if an old process is still running.

Some flavors of flock include --verbose and --conflict-exit-code args which could be used, but looks like busybox flock offers nothing.

[smlx]

It looks like alpine offers the util-linux implementation of flock with --verbose. Would just adding this package to the cli image provide enough information to address this?

[rocketeerbkw]

That would probably work (didn't test output with go-crond), but because we add this in build-deploy-tool and not the image, how do we make a change w/o breaking anyone that doesn't have the alpine flock installed?