Path-Based Authorization

Path-Based Authorization
Prev	Chapter 6. Server Configuration	Next

Both Apache and svnserve are capable of granting (or denying) permissions to users. Typically this is done over the entire repository: a user can read the repository (or not), and she can write to the repository (or not). It's also possible, however, to define finer-grained access rules. One set of users may have permission to write to a certain directory in the repository, but not others; another directory might not even be readable by all but a few special people.

Both servers use a common file format to describe these path-based access rules. In the case of Apache, one needs to load the mod_authz_svn module and then add the AuthzSVNAccessFile directive (within the httpd.conf file) pointing to your own rules-file. (For a full explanation, see the section called “Per-Directory Access Control”.) If you're using svnserve, then you need to make the authz-db variable (within svnserve.conf) point to your rules-file.

Do you really need path-based access control?

A lot of administrators setting up Subversion for the first time tend to jump into path-based access control without giving it a lot of thought. The administrator usually knows which teams of people are working on which projects, so it's easy to jump in and grant certain teams access to certain directories and not others. It seems like a natural thing, and it appeases the administrator's desire to maintain tight control of the repository.

Note, though, that there are often invisible (and visible!) costs associated with this feature. In the visible category, the server needs to do a lot more work to ensure that the user has the right to read or write each specific path; in certain situations, there's very noticeable performance loss. In the invisible category, consider the culture you're creating. Most of the time, while certain users shouldn't be committing changes to certain parts of the repository, that social contract doesn't need to be technologically enforced. Teams can sometimes spontaneously collaborate with each other; someone may want to help someone else out by committing to an area she doesn't normally work on. By preventing this sort of thing at the server level, you're setting up barriers to unexpected collaboration. You're also creating a bunch of rules that need to be maintained as projects develop, new users are added, and so on. It's a bunch of extra work to maintain.

Remember that this is a version control system! Even if somebody accidentally commits a change to something they shouldn't, it's easy to undo the change. And if a user commits to the wrong place with deliberate malice, then it's a social problem anyway, and that the problem needs to be dealt with outside of Subversion.

So before you begin restricting users' access rights, ask yourself if there's a real, honest need for this, or if it's just something that “sounds good” to an administrator. Decide whether it's worth sacrificing some server speed for, and remember that there's very little risk involved; it's bad to become dependent on technology as a crutch for social problems.^[48].

As an example to ponder, consider that the Subversion project itself has always had a notion of who is allowed to commit where, but it's always been enforced socially. This is a good model of community trust, especially for open-source projects. Of course, sometimes there are truly legitimate needs for path-based access control; within corporations, for example, certain types of data really can be sensitive, and access needs to be genuinely restricted to small groups of people.

Once your server knows where to find your rules-file, it's time to define the rules.

The syntax of the file is the same familiar one used by svnserve.conf and the runtime configuration files. Lines that start with a hash (#) are ignored. In its simplest form, each section names a repository and path within it, and the authenticated usernames are the option names within each section. The value of each option describes the user's level of access to the repository path: either r (read-only) or rw (read-write). If the user is not mentioned at all, no access is allowed.

To be more specific: the value of the section-names are either of the form [repos-name:path] or the form [path]. If you're using the SVNParentPath directive, then it's important to specify the repository names in your sections. If you omit them, then a section like [/some/dir] will match the path /some/dir in every repository. If you're using the SVNPath directive, however, then it's fine to only define paths in your sections—after all, there's only one repository.

[calc:/branches/calc/bug-142]
harry = rw
sally = r

In this first example, the user harry has full read and write access on the /branches/calc/bug-142 directory in the calc repository, but the user sally has read-only access. Any other users are blocked from accessing this directory.

Of course, permissions are inherited from parent to child directory. That means that we can specify a subdirectory with a different access policy for Sally:

[calc:/branches/calc/bug-142]
harry = rw
sally = r

# give sally write access only to the 'testing' subdir
[calc:/branches/calc/bug-142/testing]
sally = rw

Now Sally can write to the testing subdirectory of the branch, but can still only read other parts. Harry, meanwhile, continues to have complete read-write access to the whole branch.

It's also possible to explicitly deny permission to someone via inheritance rules, by setting the username variable to nothing:

[calc:/branches/calc/bug-142]
harry = rw
sally = r

[calc:/branches/calc/bug-142/secret]
harry =

In this example, Harry has read-write access to the entire bug-142 tree, but has absolutely no access at all to the secret subdirectory within it.

The thing to remember is that the most specific path always matches first. The server tries to match the path itself, and then the parent of the path, then the parent of that, and so on. The net effect is that mentioning a specific path in the accessfile will always override any permissions inherited from parent directories.

By default, nobody has any access to the repository at all. That means that if you're starting with an empty file, you'll probably want to give at least read permission to all users at the root of the repository. You can do this by using the asterisk variable (*), which means “all users”:

[/]
* = r

This is a common setup; notice that there's no repository name mentioned in the section name. This makes all repositories world readable to all users. Once all users have read-access to the repositories, you can give explicit rw permission to certain users on specific subdirectories within specific repositories.

The asterisk variable (*) is also worth special mention here: it's the only pattern which matches an anonymous user. If you've configured your server block to allow a mixture of anonymous and authenticated access, all users start out accessing anonymously. The server looks for a * value defined for the path being accessed; if it can't find one, then it demands real authentication from the client.

The access file also allows you to define whole groups of users, much like the Unix /etc/group file:

[groups]
calc-developers = harry, sally, joe
paint-developers = frank, sally, jane
everyone = harry, sally, joe, frank, sally, jane

Groups can be granted access control just like users. Distinguish them with an “at” (@) prefix:

[calc:/projects/calc]
@calc-developers = rw

[paint:/projects/paint]
@paint-developers = rw
jane = r

Groups can also be defined to contain other groups:

[groups]
calc-developers = harry, sally, joe
paint-developers = frank, sally, jane
everyone = @calc-developers, @paint-developers

Partial Readability and Checkouts

If you're using Apache as your Subversion server and have made certain subdirectories of your repository unreadable to certain users, then you need to be aware of a possible non-optimal behavior with svn checkout.

When the client requests a checkout or update over HTTP, it makes a single server request, and receives a single (often large) server response. When the server receives the request, that is the only opportunity Apache has to demand user authentication. This has some odd side-effects. For example, if a certain subdirectory of the repository is only readable by user Sally, and user Harry checks out a parent directory, his client will respond to the initial authentication challenge as Harry. As the server generates the large response, there's no way it can re-send an authentication challenge when it reaches the special subdirectory; thus the subdirectory is skipped altogether, rather than asking the user to re-authenticate as Sally at the right moment. In a similar way, if the root of the repository is anonymously world-readable, then the entire checkout will be done without authentication—again, skipping the unreadable directory, rather than asking for authentication partway through.

^[48]A common theme in this book!

Prev	Up	Next
httpd, the Apache HTTP server	Home	Supporting Multiple Repository Access Methods