If you install GIT or SVN (Server Repo) on Linux, will it run any permanent processes i.e. will it leave a permanent memory footprint on my Linux server?

I'm on a server where keeping down memory usage is of the essence.

link|improve this question
4  
do you know of any programmer who doesn't use version controls. get a life. – Adergaard Oct 28 '09 at 19:59
1  
I don't know of any programmers who don't breathe. That doesn't mean that air-related questions should be considered relevant. This belongs on serverfault. – Beska Oct 28 '09 at 20:03
4  
@Beska: clever. I give you that. but seriously, why are so many so hard on someone who needs help. There's obviously a bunch of ppl out there who knew the answer to this semi-programming-related question. If stackoverflow becomes some sort of elitist place where only this but not that can be discussed, then it's on its way out. Think about it. – Adergaard Oct 28 '09 at 20:08
Would this be a central store or just a client? – Matthew Whited Oct 28 '09 at 20:08
@Matthew Whited: well, in general I like to have a central repo where I know it won't disappear if I accidently set the building on fire and lose the hand held, laptop or home server. So central store is a must, even though I guess GIT really has no "central", it's simply where you did push or pull the latest. – Adergaard Oct 28 '09 at 20:11
show 4 more comments
feedback

migrated from stackoverflow.com Oct 29 '09 at 21:15

This question came from our site for professional and enthusiast programmers.

3 Answers

Subversion will require you to run either the svnserve daemon or mod_webdav in Apache to access your repositories. If you are looking to keep things to a minimum, you can set up svnserve to run as an inetd service (this is how I run a Subversion server on a little linux laptop in the basement) or using svn+ssh. Note that in all cases, a daemon is required to accept the incoming request, but you maybe able to piggyback on an existing service.

GIT can be used without a daemon of any kind, if you simply plan to use file sharing to synchronize your repositories. This isn't an ideal setup for a large development group, but it is workable if you are really constrained.

If one of these sounds better than the other, let me know and I can elaborate further.

link|improve this answer
superb. thanks for a fast answer. – Adergaard Oct 28 '09 at 20:00
2  
Subversion can run in a serverless mode like git does, look at svn+ssh repo designations. – Jim Zajkowski Oct 28 '09 at 20:01
Subversion doesn't require daemon if you go svn+ssh – Arkaitz Jimenez Oct 28 '09 at 20:01
feedback

Don't know for git, but for svn: you have multiple options.

If you use svnserve yes, it will have a running process. If you use svn+ssh protocol the svn process will only get created when needed, and not leave a permanent process running. If you already have an apache httpd running you can also let that one deliver svn content, in which case you have obviously the apache footprint.

link|improve this answer
With git you have the same mode as svn+ssh, it works basically the same using the opened ssh connection to pipe the protocol. – Arkaitz Jimenez Oct 28 '09 at 20:08
1  
In this case, SVN doesn't need a resident process, but a daemon is still required (sshd) to accept the request. – jheddings Oct 28 '09 at 20:10
feedback

Actually, subversion allows you to run without memory footprint, through ssh, will execute svnserve only for the moment you are commiting.

link|improve this answer
It sounds like the server is very constrained... I didn't want to assume sshd was running. If it isn't that will have to be added, in which case, a daemon is required to access SVN. – jheddings Oct 28 '09 at 20:11
2  
If you can't run sshd then you have a really tiny machine, and if that's the case, surely svnserve or git will blow the doors off. – Jim Zajkowski Oct 29 '09 at 21:28
feedback

Your Answer

 
or
required, but never shown