Child pages
  • Indore 10 EAP2 (build 41463) Release Notes

Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

The settings archive also contains a report.log file detailing the reasons for exporting external entities.

Smart checking for changes interval

Traditionally TeamCity uses polling for detecting changes in VCS repositories. Polling is highly reliable approach. It works in the majority of cases. Even if TeamCity server was stopped for a while, with polling it can easily pickup all the changes made in repositories on the next startup.

But polling also has one downside which becomes more and more important, once TeamCity installation grows. If there are many different VCS repositories configured in TeamCity, polling can impose significant load on both TeamCity server and VCS repository servers.

Alternative approach is to use push model: various post commit hooks and web hooks. This approach is more scalable, but it cannot be used alone. Because if TeamCity server is stopped, obviously all push notifications will be lost.

This is why we decided to implement combined approach. Starting with this EAP if commit hook initiates process of checking for changes for some VCS root in TeamCity, TeamCity will automatically increase checking for changes interval for this VCS root, assuming that this commit hook will now come to TeamCity on a regular basis. But, if so happens that TeamCity detects a change in this VCS root during regular polling, then checking for changes interval will be reset to initial value specified by user when VCS root was created. This is done for the case when commit hook stopped working for some reason. If TeamCity server was restarted, it will switch to polling for all of the VCS roots, till commit hooks start informing it about new commits.

Please refer to our documentation on commit hooks configuration in various VCS repositories: <link>

GitLab support in Commit Status Publisher

...