How to understand why TeamCity select certain revision for build

Hello,

recently we periodically faced with problem - TeamCity select incorrect revision for build.

TeamCity correctly calculate "upper limit revision" revision, but "Latest commit attached to build configuration" - not.

Is it possible trace logic why this revision was selected? In logs or other way?

 

More details:

 

Teamcity version: TeamCity Enterprise 2017.1.4 (build 47070)

We use Git, build run on different branches.

We have checkout rules.

In last 3 observed cases Teamcity does not select merge revision, but select revision from parent where changes (related to this build configuration) was made more recently. Actually both parents contains related changes and so, merge revision should be selected. Also it somehow related with builds history, because problem vanished if either delete all builds in history or made one more change related to build configuration.

 

I try reproduce problem in test repository, but without success. I have problem just now and can perform investigation.

0
4 comments

Hi Sergey,

how do you trigger the builds? Are they triggered manually? Or via a VCS Trigger?

0
Avatar
Permanently deleted user

Here is no VCS trigger on this particular build configuration.

Problem reproduced if build triggered manually or as part of build chain.

 

0

Hi Sergey,

do the different configurations in the build chain with that VCS Root have different branch specs? This is usually the scenario where a build chain doesn't pick the "right" revision, as it's not available to some of the build configurations, so the last one that matches is used to preserve consistency.

0
Avatar
Permanently deleted user

Not sure if I understand question correctly... When build run as part of chain - yes some builds can start on different branches:

(EDIT: on screenshot no 'bad' build in list, it is only example)

But I do not sure if it is this case, because when I run it manually - it failed also. And this build is 'leaf' (has no dependencies):

0

Please sign in to leave a comment.