I'm moving my current server contents to a new one, and am currently in the process of setting up SVN. I'm fairly unfamiliar with SVN, typically using it to the extent of commits and updates.
I have two locations that I use SVN on the old server:
PROD location:
/var/www/html/new_dwutils/
and local:
/home/{user_name}/public_html/new_dwutils/
My interaction svn-wise is normally committing and updating at the /new_dwutils/ level.
Note: Running svn --version says I'm at version 1.6.11 for both servers.
I'm now trying to recreate this structure on the new server. My initial thought was to create the svn repo using something like:
svnadmin create /var/www/html/new_dwutils/
This creates the repo dir, but, when I copy my files into the dir, I am unable to do svn commands like status. However, when I go into a sub-dir of the copied data, I can use the svn commands.
This has me thinking that the repo is /new_dwutils/ and the copied data is considered a project? And the sub-dirs are working copies then?
Going off that thought, I deleted the repo, and made the html dir a repo:
svnadmin create /var/www/html/
I then copied my new_dwutils dir, and sure enough, I was able to do svn commands like I use too. What I've noticed is that when creating the repo, a few things are added that were not on the previous server: conf/, db/, format, hooks/, locks/, and README.txt. I get that these are svn files, but I'm not seeing the .svn file. I know that there was an update for svn that "removed" .svn files, but these files are now in /var/www/html/.
Now I want to setup my local working copy.
I've been doing (location /home/{user_name}/public_html/):
svn checkout file:///var/www/html/
Problem is it copies the html/ file, but nothing in it, and I don't want the html/ file I want the html/new_dwutils/ file.
I feel like I'm doing it wrong from the start, and would greatly appreciate some explanation on how to get on the right track. A step by step would be extremely useful, and if further clarification is need for files or directory paths, I would gladly detail.
Thanks!
The Subversion Manual will answer all of your questions.
If you're making a Subversion repository under /var/www.html, I'm assuming you're using Apache httpd as your server. Look at Chapter 6. If you already have a repo, create a dump file, then use that dump file to recreate the repo. Look at Chapter 5 on moving repositories.
If you don't know anything about Subversion, or are confused by the difference between the repository location directory and a working directory, read the on-line manual. It's one of the best pieces of documentation I've seen.
From description of your question it appears that '/var/www/html/new_dwutils/' is your working copy and not a repo.
Go to '/var/www/html/new_dwutils/' on the old server and type "svn info" this should give you location of the old repo. You should simply be able to 'svn co ' into the new location to checkout a copy of all your files from the old server (everything that is checked in - you will not get anything that is not checked in on the old server).
However, if your repo was local on the old server and you want to move it to your new server too. Then you can simply copy the entire folder to the new server and access it directly using its new location in 'svn co' command.
Related
I spent all day installing the svn server on my virtual debian 7. So now I have working svn server on svn:// and http:// protocols. With the http:// I haven't got any problems, but with the svn:// first I couldn't create a project folder in my repo through Tortoise browser, there was some error: "User cancelled", after this I did it in the console with some command like this one
svn mkdir -m "project folder" svn://svn.host/repo/project
then a message for root password appeared, next svn user name and pass, and the project folder showed up to the Tortoise browser. I did 3 folders trough Tortoise browser: trunk, branches and tags, and I added my local project to the trunk directory. So now everything seems to work fine just when access my repo through console, ftp or http I can see the trunk branches and tags directories. I tried to find them in the linux tree with no success, it is very strange because the project is there in the Tortoise Browser, but not in the repo. Thanks of any good people which can help me to understand whats going on, and sorry for the level of my English still learning.
The subversion repository is basically a database; you won't see the literal names of files and directories if you inspect it directly. This follows from the purpose of subversion: to store the history of every file and folder, including its state at every past revision, and metadata such as log messages, and actions which change the directory structure, such as renames and deletes.
Laying out the repository the same way as the code being stored would be a very inefficient solution for this aim.
I created a new project in svn svnadmin create /myrepo in my server, with my client I did a checkout and add new files, later a commit, so, if I make a checkout from another computer I get the recently added files, which is perfect, but at my folder /myrepo still is no file, all the new files that were added from my client are not visible there, I know they implement many algoritms to take the version control, my question is, should I be able to see all the new files added from /myrepo in my server, without need to make a checkout with a client or something like that??
I want to know where my files are saved at my server,
Thanks
No. The files are stored in the repository you created, but in a specialized database. If you go to myrepo and look in the db folder, you'll see that there are revision files stored there. Those files contain the structure and data of the repository at specific instances in time. The Subversion book has some information on the structure. You can also look at the documentation in the actual Subversion repository about the structure used to store the data.
I am wondering what happens in SVN when a file is updated directly instead of using SVN? The main reason I am asking is that there was a problem updating the SVN on my machine (windows) when the server (linux) had 2 names that were the same, but different case. I resolved this on the server, but didn't do it through SVN since it won't update correct, but I still get the issue. Do I need to run some kind of command to update it?
Thanks.
EDIT:
I deleted the comflicting file in the working direcotry and wanted to know if doing things directory in the working directory get tracked at all or what needs to be done to resync.
When SVN gets blocked because the repository is more "up to date" than the local checkout, one brain dead foolproof solution is:
Move (or remove) the files that are causing the conflict at the command line (don't use SVN tools to do this, and don't use the GUI if you have tortoise installed).
svn update the repository, which will restore the current copy of the files from the subversion server.
Decide what to do with your cached copies of the old files. Either manually merge them back into the repository, discard them, or remake the changes in the new svn managed files (depending on your needs).
Note that if you move the files into a directory using tortoise, make sure that you move it into a directory that's not associated with ANY SVN project. It's not fun trying to undo the helpful changes tortoise does in thinking your wanting a SVN move to accompany the file system move.
There is no need to run any special commands. If you updated the sources, the next time you will run svn update subversion will seamlessly merge the changes and you will get an uptodate working copy.
If you changed some files, they will appear modified or conflicted depending on the changes made by you and other users.
I have been going through documentation and such and have SVN working, but I want to put it on an existing directory. I imported that directory, so do I rename/delete the non SVN directory and then checkout the SVN to the non SVN directory location? I am just trying to understand how to get it to start posting to our website URL.
If so, is there any way to keep the current non SVN and make it SVN rather than import and overwrite?
Thanks, I am trying to understand SVN, but find a lot of the tutorials and such on the web to be confusing.
Yes, you have it exactly. Once your code has been added to the repository, you can get rid of or rename your original code directory. Then checkout the project from the repository into the same location as your previous code and continue working from there.
UPDATE
To make it so that your website is updated from the repository, you actually need two working directories, and a repository.
Repository: The repository stores the code and changesets, but isn't directly accessible as a file system. Keep a backup!.
Working directory 1: You develop and test your code from a working directory checked out from the repository. Commit changes back to the repository.
Working directory 2: Rename the code directory on your webserver. Checkout a copy of the code to your web server in its place. Technically it is now a working directory, since it contains the .svn metadata directories, though you won't usually make changes here.
Make changes to your code from your development working directory (1) and commit them back to the repository. When you are satisfied that they are working correctly and have been properly tested, on the web server's code copy (2) do svn update (or if you're using Tortoise SVN on the web server, do an update). This will synchronize the server code with the current development version.
Subversion will not automatically push updates to your web server. You will need to pull them in with an update when you need to. It is possible to use what's called a "post-commit hook" to cause Subversion to execute a script when commits are made, and that script could update or export code to your production web server. However, you would need to write the script and it's kind of an advanced usage of Subversion. I would recommend trying out the method I described with a working copy on the web server to get accustomed to the workflow befrore trying anything more complicated.
Addendum If you really want to do this (and I don't really recommend it unless you really test well) a very easy method would be to schedule a cron job that does svn update every couple of hours (or minutes) on your production site.
Don't forget that if you do happen to modify your code directly on the web server, you must commit it back to the repository from there, and do an update on your development working copy.
I want to add my project to a subversion repository. The project folder contains a symlink to a folder containing thousands of txt files that I don't need to add to the svn repository. I DO want the symlink-folder to show up when I checkout the code, however.
It looks like I can use svn addprop svn:ignore symlinked-folder to ignore the folder, but then I'll have to add that symlinked folder to every working copy I check out before everything will work.
Is there a proper way to do this?
Perhaps there is no way to deal with this, since a symlink is a filesystem artifact. Is there a better way to handle this situation?
CONCLUSION - EDIT
After all this investigation, I committed the symlink-folder by accident and SVN added it to the repository without adding any of the files within it. Upon checkout, everything works fine. The symlink-folder checked out and works.
I am using assembla to manage my SVN repository, so that might have something to do with this success.
The answers above are right, your symlink won't work if you check out the repository on windows.
But if you're aware of that and you don't care, you can add just the symlink without its contents:
svn add -N your-symlink
man svn add here
I believe you are correct, imagine if a user checked out your repo under Windows - how would SVN create the symlink when the underlying OS doesn't support it?
Is the target folder that you are symlinking to under version control? If so, you can make use of the svn-externals property.
You are right, it doesn't make sense to add a symlink to a repository. What would happen if someone checked out the source on a machine that didn't have access to the folder the symlink points to?
One way is to structure your repository so that you can check out the codebase without having to check out documents. E.g.:
Trunk
Tags
Branches
Documents
So you only check out the trunk or branch that you are working on, and when you require it you can check out the documents.
Alternatively, use a project management tool like Redmine to store your docs. It can integrate with svn as well so you can view your repository and manage permissions through it.