My server runs Redhat 9. One client that I host would also like me to host CVS for him. I don't have any experience in working with CVS, or installing it.
Sorry I have not read all the docs and i don't know all the proper terms. I am looking for some tips so I can make the decision on whether it is worth it to proceed. (I don't want to waste my time if it isn't)
Ideally I would like to be able to use the CVS as well; but not allow my client to view/download my projects. --projects probably isn't the correct term -- My client should be able to create projects also but not be able to access/modify mine.
What kind of issues are there with performance?.. CPU usage, and memory requirments?
Is it ok to run a CVS server along with the webserver? Or would is be better (more secure?) if the CVS was on a separate server?
How can I install CVS and maintain these security restrictions? Any good online tutorials?
duh... i know the cvs homepage. I am not looking for that. I am looking for people's experiences and reccomendations. I could spend hours looking this stuff up but I just don't have time to do that if this is not somthing I should pursue. I'm not lazy... I just prioritize.
CVS isn't too bad to install and make somewhat secure. Basically you'll have any number of "roots" in which you can set different logins and passwords. This would allow you to have your code separate from his code. You may or may not want the CVSROOT for a particular branch to be in a users web space. If the user wants to dynamically modify their site by checking in a new html file then it may make sense to have cvs and your webserver share a directory tree. This certainly would not be the case for more important files like the webserver configuration and so on.
From a performance perspective I'd say that a CVS server is not particularly resource intensive. Normally it will only be run on demand by xinetd though you could configure it to be running all the time. The protocol allows for compression during communication so that you won't take as much bandwidth. Of course, having said that if the user decides to check in binary images of the Linux kernel or some such then, yes, there will be a bandwidth and I/O hits. Normal source code should not be too bad though.