Hi,
I don't know if this situation could be happening:
- Hi SCM! I'm a Jenkins job and last time I polled was 000000000000000000 SHA-1 hash. Do you have new content for me?
- Yes, I have new content and new tags from that content.
- Ok, see you, I have a lot of work to do, because I must execute my build steps.
- Wait! Don't forget your new SHA-1 hash 111111111111111111.
...
- Hi SCM! I'm a Jenkins job and last time I polled was 111111111111111111 SHA-1 hash. Do you have new content for me?
- Yes, I have new content, but no tags.
-
Oh! So, I have nothing to do. Ok, see you, I gonna take a
break.
- Wait! Don't forget your new SHA-1 hash 22222222222222222.
...
- Hi SCM! I'm a Jenkins job and last time I polled was 22222222222222222 SHA-1 hash. Do you have new content for me?
- Yes, I have new tags, but they are referring to previous content.
- Oh! Tags referring to previous content! So, I have nothing to do, because I'm looking for tags referring to commits from my last SHA-1 hash. Ok, see you, I gonna take a break.
- Wait! Don't forget your new SHA-1 hash 333333333333333333.
...
I mean, were there polling executions between committed changes and tag creation referring to some of those commits? If you'd execute polling once at night, and committed changes and tag creation referring to some of those commits (not HEAD) would have happened on the same day, would you have had this problem?
It's a pity you can't configure hooks, because it seems the best solution.
I hope be useful.
Best regards,
Rachel