How does the "Enable repository caching on remote agents" work?

[Maggie Stearns]

I have a question about the Bamboo feature that can be set when defining a git repo from Stash called “Enable repository caching on remote agents”.

 Can you give me some additional information on his this caching works? Does the main Bamboo server somehow maintain a cache that the agents use? or does each agent get a clone of the git repo and act as a cache for all other builds on that agent?  Just wondering how to describe the benefits of this feature.

[Przemek Bruski]

When cache is enabled, if you have multiple builds (including, but not limited to branch builds) that use the same repo, the initial full fetch of the repo will happen only once per remote agent. Otherwise each build will issue a separate remote fetch. 

[Maggie Stearns]

So does a copy of the git repo sit on the Bamboo server?

[Przemek Bruski]

This option is not related to server in any way. The server always caches repos.

[Marcin Gardias]

The remote agent cache sits in the respective agent's home. Server cache is not controlled by this option but it's only used by server-side work and local agents.

[Mike Friedrich]

Yes,  it is a git repo on the server and on each agent. 

[Maggie Stearns]

So when a second agent wants to build the same set of code, it is faster because of the cache on the server?

[Maggie Stearns]

Does my Bamboo server now need to store a copy of the git repo? (where it was not storing it before?)

[Marcin Gardias]

"So when a second agent wants to build the same set of code"

No, assuming "second agent" is a new remote agent, it will need to do an initial fetch. Subsequent builds will be faster though (at the cost of storage space)

"Does my Bamboo server now need to store a copy of the git repo?"

It always did, no change there, the option only refers to remote agents.