I have the latest version of TeamCity (as of this posting date) installed on OSX (leopard), projects and are all setup and running, I have an agent running on the same machine, it checks out my repository fine (agent checkout) and executes builds.
On the same osx machine I have SVNServe running (as SvnServe daemon, not apache or inetd). Initially I was using 1.5.5, but I've upgraded the server to 1.6.1 (IIRC - currently away from machine) and I still have the same problem...
Whenever teamcity checks for changes svnserve seems to leave a process behind and eventually when I reach about 200 threads/processes teamcity fails because SVNServe won't allow any more handles to my repository files to be opened.(the file handle limit is currently 256 on my mac)
I was running svnserve happily on a linkstation, but when I started using teamcity that would eventually cause a crash and shutdown the system. I never tracked that particular issue down, but I can imagine it was a similar problem running out of memory/file handles etc.
I've been researching on google and the only thing I can find is that it might be the fault of Teamcity not sending the correct EOF signal to svn. I don't know the protocol, but this seems a reasonable fault of teamcity's...
I've not tested thoroughly, but SvnX and tortoise don't seem to be leaving behind processes when I checkout or check logs, only TeamCity.
Has anyone had similar experiences? Does anyone know of any bugs running teamcity, an agent, or SvnServe on OSX?
Is there anywhere deeper I can check?
 Same applies for checkouts etc, but I can just leave teamcity server checking for changes and it'll eventually fail.
 I've tried switching svnserve to be threaded and it's exactly the same behaviour