[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: CVS Unix to Linux Migration

From: Todd Denniston
Subject: Re: CVS Unix to Linux Migration
Date: Wed, 04 Feb 2009 09:58:47 -0500
User-agent: Thunderbird (X11/20081209)

Rez wrote, On 02/04/2009 12:49 AM:
Hi Everyone
I'm in the midst of migrating our current repository from an old Solaris box to a new Redhat(CentOS 5.2) linux box. CVS is installed, configured, and all set up on the new server. Users have been re-created and setup in /etc/passwd. I created a test Repo and from a Windows client machine using WinCVS I managed to connect via the pserver method and checkout a project/module successfully.

I assume this means you have modified the /etc/xinet.d/cvs file correctly and thus inet recognizes calls to it on 2401.

Could someone please tell me: 1- if the migration is more involved than simply tarballing the repository from the old server and untarring and mounting it on the new server? Meaning, the repository is independent and not affected by the old OS in any way as far as file system or formatting or any other thing go.

The file structure should be good.
The permissions/ownership may need to be tweaked if all the names/UID/GID of the users do not match from system to system.

 What else do I need to do on the old server to prepare?
On the old server, I would disable cvs in /etc/inet or /etc/network/inet (where ever the Solaris you are working with hid it's inetd config) and restart inetd... BEFORE making the final tarball to put on the new machine. Reason: you don't want to loose any changes someone makes while you are turning on the new machine.

2- Because it's a migration by way of untarring, do I still need to execute "cvs -d /repo/path init" since the existing repo already contains the CVSROOT directory?

It is still a good idea, because by doing that cvs will create, with default settings, any new config files that did not exist when the old cvs was made.

3- Also, I would like to get rid of some old projects in the repository before I migrate it, we don't need the history and don't need to save them, so could I just log into the old server as Admin and do an rm or mv command (carefully of course) w/o trashing or corrupting the repository?

rm or mv in the repo is by definition "corrupting the repository". :)
I would on the new server, build a script that did appropriate rm's based on where you are putting the final repo and what you know needs to go away, then when you untar the last tarball, run the script on the new repository. This way, if you quickly figure out you made a mistake, you still have everything as it was on the old server.
Summary: keep the old server as it was, so it is a back up to the backup. :)

Thanks all Rez

Todd Denniston
Crane Division, Naval Surface Warfare Center (NSWC Crane)
Harnessing the Power of Technology for the Warfighter

reply via email to

[Prev in Thread] Current Thread [Next in Thread]