Reducing agent polling interval?

We're currently building a virtualized acceptance testing environment on top of TeamCity and HyperV. The actual test cases are run on virtual machines which are reset to snapshots before each test run.

The basic infrastructure seems to be working nicely, but there's a gap of several minutes between a virtual machine and its TC agent being restarted and the agent being reconnected to the server.

It seems to me, by looking at the agent console log, that the gap is caused by the server rather than the agent itself. Apparently the server only polls the agent(s) every few minutes. Is there a way to reduce this interval to, say, every 30 seconds?

It's not a huge issue at the moment, but might be getting bigger as we start adding more (virtualized) base configurations to the setup. I tried searching for configuration options/startup properties and related issues and discussions, but couldn't find anything.

2 comments
Comment actions Permalink

Eetu,

When you start (or restart) the agent process, it should connect to TeamCity server immediately.
But at first start the agent automatically downloads and installs a set of plugins from the server, and it takes several times.
Please check upgrade.log file, do you see any activities there in this moment?

0
Comment actions Permalink

No, there's nothing in the upgrade.log after the initial startup of the agent.

However, I did spot a couple of instances in teamcity-agent.log where the client seems to have lost the connection to the server immediately after establishing it. It then takes some time before the connection is re-established. I'll have to see whether the problem persists and if it does, how often it occurs.

0

Please sign in to leave a comment.