From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3a) Gecko/20021212 Description of problem: I just installed rh8.0. A standard CVS setup works OK to localhost but when I try to check something out over ethernet, both client and server hang. The client is cvs-1.11.1p1-3. A command like cvs co dbench get stuck, client is waiting or TCP data (stuck in tcp_data_wait) and the server is also in interruptible sleep. Looks like protocol failure and each end is waiting for the other. downgrading the cvs on the server (rh8.0) to cvs-1.11.1p1-7 fixed it up. And upgrading the client from cvs-1.11.1p1-3 to cvs-1.11.2-5.i386 fixed it up too. So it appears that cvs-1.11.1p1-3 and cvs-1.11.2-5.i386 will not talk to each other. Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. See above 2. 3. Additional info: See above
I believe I'm having the same issue; I can't get the user to tell me the client info at the moment, but the symptoms are the same: the client just stops in the middle of a checkout; both sides will wait indefinitely for in a read. Unfortunately I can't duplicate it on any of the machines I have access to. To be fair, I am running the cvs-1.11.2-5 RPM with one extra patch (to export CVS_Username to an environment variable) rebuilt on a 7.2 machine. None of this should make any difference to the networking, however.
I probably experience the same problem seen on RedHat 9 cvs-1.11.2-10.i386.rpm. Upgrade to rawhide cvs-1.11.5-2 resolved the problem for me, rebuilt from source as no binary was available in the rawhide mirror I use... In my case the problem is compression (-z) which is broken for older CVS clients in 1.11.2 (see cvs 1.11.3 release notes), but have also seen much worse effects (100% CPU usage until manually killed cvs server processes) in another installation of 1.11.2.