Closed funkyfuture closed 4 years ago
Looks like idempotence is broken by this PR
i'll probably not have the time too look into this within the next two weeks.
the thing is, from the logs i can't gather how the break of idempotence is detected as a i have the impression that the playbook is applied only once. if you know a good resource that conveys the gist of how these checks work, i'd appreciate if you pointed it out. also, i'm puzzled why Docker is involved in the testing setup.
The error message is quite clear: https://travis-ci.com/github/systemli/ansible-role-jitsi-meet/builds/164238339#L609
I recommend to do the local testing with molecule-vagrant
and then use molecule converge
multiple times. It shows idempotence errors quite clearly.
a'ight i fixed this by using the base secret as salt to the hash function that would then turn to be deterministic.
i simplified the password derivation chain. i guess i got hooked on the password_hash
filter b/c i initially wanted to feed the base secret as salt to it.
Thank you. I think this is an amazing feature and smart implementation!
this is a follow-up w/ only the proposal for a base secret as follow-up to #25.