SCM works perfect over https but not http?

With the default configuration of SCM which only supports HTTP, we encountered regular problems with all repos such as the inability to checkout a branch tag or trunk without errors (either timeout errors or REPORT request on ‘/!svn/vcc/default’ failed) even though checking out the entire repo worked every time. The logs showed that the request was handled properly and that a timeout occured 30 seconds after (assuming because the client stopped talking).

However, when we implemented a self-signed HTTPS in the configuration and used this instead, all of those problems went away. No more timeouts or REPORT errors on the clients.

Has anyone heard of this behaviour before?

Hello @pcc2,

This is the first time we have heard of such a problem. We were wondering if you are using a reverse proxy? Also could you provide server logs of these errors?

Regards, Eduard

There is no reverse proxy that we know off, assuming there’s something on the network however that interferes with some traffic and not other if you have not known about it before. Perhaps traffic shaping or some such that when encrypted leaves the traffic alone? The network engineer can’t see anything wrong and is always busy so will leave it with https for now as no problems since swapping to it.