Previous: statusgui, Up: Developer Tools

5.2.3 try

This lets a developer to ask the question “What would happen if I committed this patch right now?”. It runs the unit test suite (across multiple build platforms) on the developer's current code, allowing them to make sure they will not break the tree when they finally commit their changes.

The buildbot try command is meant to be run from within a developer's local tree, and starts by figuring out the base revision of that tree (what revision was current the last time the tree was updated), and a patch that can be applied to that revision of the tree to make it match the developer's copy. This (revision, patch) pair is then sent to the buildmaster, which runs a build with that SourceStamp. If you want, the tool will emit status messages as the builds run, and will not terminate until the first failure has been detected (or the last success).

There is an alternate form which accepts a pre-made patch file (typically the output of a command like 'svn diff'). This “–diff” form does not require a local tree to run from. See See try –diff.

For this command to work, several pieces must be in place:


The buildmaster must have a scheduler.Try instance in the config file's c['schedulers'] list. This lets the administrator control who may initiate these “trial” builds, which branches are eligible for trial builds, and which Builders should be used for them.

The TryScheduler has various means to accept build requests: all of them enforce more security than the usual buildmaster ports do. Any source code being built can be used to compromise the buildslave accounts, but in general that code must be checked out from the VC repository first, so only people with commit privileges can get control of the buildslaves. The usual force-build control channels can waste buildslave time but do not allow arbitrary commands to be executed by people who don't have those commit privileges. However, the source code patch that is provided with the trial build does not have to go through the VC system first, so it is important to make sure these builds cannot be abused by a non-committer to acquire as much control over the buildslaves as a committer has. Ideally, only developers who have commit access to the VC repository would be able to start trial builds, but unfortunately the buildmaster does not, in general, have access to VC system's user list.

As a result, the TryScheduler requires a bit more configuration. There are currently two ways to set this up:

jobdir (ssh)
This approach creates a command queue directory, called the “jobdir”, in the buildmaster's working directory. The buildmaster admin sets the ownership and permissions of this directory to only grant write access to the desired set of developers, all of whom must have accounts on the machine. The buildbot try command creates a special file containing the source stamp information and drops it in the jobdir, just like a standard maildir. When the buildmaster notices the new file, it unpacks the information inside and starts the builds.

The config file entries used by 'buildbot try' either specify a local queuedir (for which write and mv are used) or a remote one (using scp and ssh).

The advantage of this scheme is that it is quite secure, the disadvantage is that it requires fiddling outside the buildmaster config (to set the permissions on the jobdir correctly). If the buildmaster machine happens to also house the VC repository, then it can be fairly easy to keep the VC userlist in sync with the trial-build userlist. If they are on different machines, this will be much more of a hassle. It may also involve granting developer accounts on a machine that would not otherwise require them.

To implement this, the buildslave invokes 'ssh -l username host buildbot tryserver ARGS', passing the patch contents over stdin. The arguments must include the inlet directory and the revision information.

user+password (PB)
In this approach, each developer gets a username/password pair, which are all listed in the buildmaster's configuration file. When the developer runs buildbot try, their machine connects to the buildmaster via PB and authenticates themselves using that username and password, then sends a PB command to start the trial build.

The advantage of this scheme is that the entire configuration is performed inside the buildmaster's config file. The disadvantages are that it is less secure (while the “cred” authentication system does not expose the password in plaintext over the wire, it does not offer most of the other security properties that SSH does). In addition, the buildmaster admin is responsible for maintaining the username/password list, adding and deleting entries as developers come and go.

For example, to set up the “jobdir” style of trial build, using a command queue directory of MASTERDIR/jobdir (and assuming that all your project developers were members of the developers unix group), you would first create that directory (with mkdir MASTERDIR/jobdir MASTERDIR/jobdir/new MASTERDIR/jobdir/cur MASTERDIR/jobdir/tmp; chgrp developers MASTERDIR/jobdir MASTERDIR/jobdir/*; chmod g+rwx,o-rwx MASTERDIR/jobdir MASTERDIR/jobdir/*), and then use the following scheduler in the buildmaster's config file:

     from buildbot.scheduler import Try_Jobdir
     s = Try_Jobdir("try1", ["full-linux", "full-netbsd", "full-OSX"],
     c['schedulers'] = [s]

Note that you must create the jobdir before telling the buildmaster to use this configuration, otherwise you will get an error. Also remember that the buildmaster must be able to read and write to the jobdir as well. Be sure to watch the twistd.log file (see Logfiles) as you start using the jobdir, to make sure the buildmaster is happy with it.

To use the username/password form of authentication, create a Try_Userpass instance instead. It takes the same builderNames argument as the Try_Jobdir form, but accepts an addtional port argument (to specify the TCP port to listen on) and a userpass list of username/password pairs to accept. Remember to use good passwords for this: the security of the buildslave accounts depends upon it:

     from buildbot.scheduler import Try_Userpass
     s = Try_Userpass("try2", ["full-linux", "full-netbsd", "full-OSX"],
                      port=8031, userpass=[("alice","pw1"), ("bob", "pw2")] )
     c['schedulers'] = [s]

Like most places in the buildbot, the port argument takes a strports specification. See twisted.application.strports for details.

locating the master

The try command needs to be told how to connect to the TryScheduler, and must know which of the authentication approaches described above is in use by the buildmaster. You specify the approach by using --connect=ssh or --connect=pb (or try_connect = 'ssh' or try_connect = 'pb' in .buildbot/options).

For the PB approach, the command must be given a --master argument (in the form HOST:PORT) that points to TCP port that you picked in the Try_Userpass scheduler. It also takes a --username and --passwd pair of arguments that match one of the entries in the buildmaster's userpass list. These arguments can also be provided as try_master, try_username, and try_password entries in the .buildbot/options file.

For the SSH approach, the command must be given --tryhost, --username, and optionally --password (TODO: really?) to get to the buildmaster host. It must also be given --trydir, which points to the inlet directory configured above. The trydir can be relative to the user's home directory, but most of the time you will use an explicit path like ~buildbot/project/trydir. These arguments can be provided in .buildbot/options as try_host, try_username, try_password, and try_dir.

In addition, the SSH approach needs to connect to a PBListener status port, so it can retrieve and report the results of the build (the PB approach uses the existing connection to retrieve status information, so this step is not necessary). This requires a --master argument, or a masterstatus entry in .buildbot/options, in the form of a HOSTNAME:PORT string.

choosing the Builders

A trial build is performed on multiple Builders at the same time, and the developer gets to choose which Builders are used (limited to a set selected by the buildmaster admin with the TryScheduler's builderNames= argument). The set you choose will depend upon what your goals are: if you are concerned about cross-platform compatibility, you should use multiple Builders, one from each platform of interest. You might use just one builder if that platform has libraries or other facilities that allow better test coverage than what you can accomplish on your own machine, or faster test runs.

The set of Builders to use can be specified with multiple --builder arguments on the command line. It can also be specified with a single try_builders option in .buildbot/options that uses a list of strings to specify all the Builder names:

     try_builders = ["full-OSX", "full-win32", "full-linux"]

If you are using the PB approach, you can get the names of the builders that are configured for the try scheduler using the get-builder-names argument:

     buildbot try --get-builder-names --connect=pb --master=... --username=... --passwd=...

specifying the VC system

The try command also needs to know how to take the developer's current tree and extract the (revision, patch) source-stamp pair. Each VC system uses a different process, so you start by telling the try command which VC system you are using, with an argument like --vc=cvs or --vc=tla. This can also be provided as try_vc in .buildbot/options.

The following names are recognized: cvs svn baz tla bzr hg darcs git p4

finding the top of the tree

Some VC systems (notably CVS and SVN) track each directory more-or-less independently, which means the try command needs to move up to the top of the project tree before it will be able to construct a proper full-tree patch. To accomplish this, the try command will crawl up through the parent directories until it finds a marker file. The default name for this marker file is .buildbot-top, so when you are using CVS or SVN you should touch .buildbot-top from the top of your tree before running buildbot try. Alternatively, you can use a filename like ChangeLog or README, since many projects put one of these files in their top-most directory (and nowhere else). To set this filename, use --try-topfile=ChangeLog, or set it in the options file with try_topfile = 'ChangeLog'.

You can also manually set the top of the tree with --try-topdir=~/trees/mytree, or try_topdir = '~/trees/mytree'. If you use try_topdir, in a .buildbot/options file, you will need a separate options file for each tree you use, so it may be more convenient to use the try_topfile approach instead.

Other VC systems which work on full projects instead of individual directories (tla, baz, darcs, monotone, mercurial, git) do not require try to know the top directory, so the --try-topfile and --try-topdir arguments will be ignored.

If the try command cannot find the top directory, it will abort with an error message.

determining the branch name

Some VC systems record the branch information in a way that “try” can locate it, in particular Arch (both tla and baz). For the others, if you are using something other than the default branch, you will have to tell the buildbot which branch your tree is using. You can do this with either the --branch argument, or a try_branch entry in the .buildbot/options file.

determining the revision and patch

Each VC system has a separate approach for determining the tree's base revision and computing a patch.

try pretends that the tree is up to date. It converts the current time into a -D time specification, uses it as the base revision, and computes the diff between the upstream tree as of that point in time versus the current contents. This works, more or less, but requires that the local clock be in reasonably good sync with the repository.
try does a svn status -u to find the latest repository revision number (emitted on the last line in the “Status against revision: NN” message). It then performs an svn diff -rNN to find out how your tree differs from the repository version, and sends the resulting patch to the buildmaster. If your tree is not up to date, this will result in the “try” tree being created with the latest revision, then backwards patches applied to bring it “back” to the version you actually checked out (plus your actual code changes), but this will still result in the correct tree being used for the build.
try does a baz tree-id to determine the fully-qualified version and patch identifier for the tree (ARCHIVE/VERSION–patch-NN), and uses the VERSION–patch-NN component as the base revision. It then does a baz diff to obtain the patch.
try does a tla tree-version to get the fully-qualified version identifier (ARCHIVE/VERSION), then takes the first line of tla logs --reverse to figure out the base revision. Then it does tla changes --diffs to obtain the patch.
try does a bzr revision-info to find the base revision, then a bzr diff -r$base.. to obtain the patch.
hg identify --debug emits the full revision id (as opposed to the common 12-char truncated) which is a SHA1 hash of the current revision's contents. This is used as the base revision. hg diff then provides the patch relative to that revision. For try to work, your working directory must only have patches that are available from the same remotely-available repository that the build process' source.Mercurial will use.
try does a p4 changes -m1 ... to determine the latest changelist and implicitly assumes that the local tree is synched to this revision. This is followed by a p4 diff -du to obtain the patch. A p4 patch differs sligtly from a normal diff. It contains full depot paths and must be converted to paths relative to the branch top. To convert the following restriction is imposed. The p4base (see see P4Source) is assumed to be //depot
try does a darcs changes --context to find the list of all patches back to and including the last tag that was made. This text file (plus the location of a repository that contains all these patches) is sufficient to re-create the tree. Therefore the contents of this “context” file are the revision stamp for a Darcs-controlled source tree. It then does a darcs diff -u to compute the patch relative to that revision.
git branch -v lists all the branches available in the local repository along with the revision ID it points to and a short summary of the last commit. The line containing the currently checked out branch begins with '* ' (star and space) while all the others start with ' ' (two spaces). try scans for this line and extracts the branch name and revision from it. Then it generates a diff against the base revision.

waiting for results

If you provide the --wait option (or try_wait = True in .buildbot/options), the buildbot try command will wait until your changes have either been proven good or bad before exiting. Unless you use the --quiet option (or try_quiet=True), it will emit a progress message every 60 seconds until the builds have completed.