Mercurial - High CPU Usage when Pushing Changes to Server Repository

We recently implemented Mercurial within one of our teams at our company as a trial before rolling it out to everyone. So far, everything has gone pretty good. But when we let another team access to Mercurial, we ran into an issue when pushing changes from our local repository to the server repository.

The python.exe process is using close to 100% CPU on the server when a push is in progress. And this is wit开发者_JAVA技巧h one single push going on. It was at originally at 100% but we added server.uncompressed = true in the hgrc and it seemed a help a little bit but it's still high.

The server is a VM running Windows Server 2008 Standard on an Intel Xeon 3 GHz with 2 GB RAM.

Doing a Google search yielded no useful information. Does SO have any suggestions?

---

Are you really sure there's a problem somewhere ?

I really don't see what's wrong with using all the CPU power at disposition to do the job.

Using compression probably "helped" because python has to access to wait for the decompression library to do its work.

Do you prefer to wait for 5 seconds at 20% load, or 1 second at 100% load ?

---

You might be running into issue #135. Try hosting the repository over https instead of accessing it via ssh.