2 Bacula Regression Suite and CTest
3 ==================================================================
5 Thanks to Frank Sweetser, the Bacula regression scripts have now been modified
6 to use the ctest component of cmake. The major gain from this, since Bacula
7 already had a working test framework in place, is the ability to have the
8 results of each test submitted to a centralized dashboard system. All of the
9 test results are aggregated and summarized, where all of the developers can
10 quickly see how the regression tests are running.
13 ==================================================================
15 For more complete documentation on ctest, go to:
17 http://www.cmake.org/Wiki/CMake#CTest
19 The first step is to install the cmake package, which includes ctest. If your
20 distribution does already package it, you can download it directly from the
25 Next, you must edit your regression config file and add a paramter called
26 SITE_NAME to identify the machine running the tests. Ideally, it should
27 contain something to identify yourself to whoever is viewing the test results
28 as well as something to allow you to identify which machine is running the
31 SITE_NAME=kern-bacula-gumbie
33 Once you have cmake installed, you can perform one of two different kinds of
34 runs to submit test results. The most common kind will be Nightly runs. A
35 Nightly CTest backup will update the source directory (as defined by the
36 BACULA_SOURCE setting in your config file) to the current version, run the
37 specified list of tests, and submit all of the results to the server. Note
38 that all of the results in a given 24 hour period (starting at 9pm EST) are
39 lumped together to appear as a single block, rather than each test showing up
42 The simplest way to trigger a nightly run is to use one of the two provided
43 scripts. The nightly-all script will run all non root tests, both tape and
44 disk based, while the nightly-disk script will run only the disk based tests.
45 So, you can choose between the following scripts:
48 ./nightly-all # does disk and tape testing
49 ./nightly-disk # disk only tests
51 ./experimental-all # experimental disk and tape testing
52 ./experimental-disk # experimental disk testing
54 We recommend that you start with the ./experimental-disk runs so that
55 you can check that everything is working fine. Once that is done,
56 try a nightly-xxx run. The difference is the experimental runs are just
57 that -- they are things where you are experimenting and it is expected that
58 something might be broken (bad ctest configuration, experimental source
59 code, ...), and nightly runs are not expected to fail.
61 If you are a developer and you have modified your local SVN repository, you
62 should be running the experimental tests -- they are designed for developers.
63 If you do modify your local repository and commit it, then run a nightly
64 test, the local repository may be reverted to a prior version so that the
65 nightly tests all have a consistent cutoff time.
67 If you are just doing testing on a nightly basis (no development in your
68 source repository), then please use the nightly tests.
70 All the old scripts (./do_all, do_file, all-non-root-tests, ...) manually
71 run the tests outside of ctest.
73 Periodically, however, you may want to submit a single test separately from a
74 weekly run. This may be a test of a particular patch you're working on, or
75 perhaps a new OS patch. For these one-shot tests, you will want to manually
76 run ctest in Experimental mode, something like:
78 REGRESS_DEBUG=1 ctest -D Experimental -R all-non-root:auto-label-test
80 The '-D Experimental' option tells ctest to submit the test results as
81 Experimental instead of Nightly. We reccomend you use the REGRESS_DEBUG
82 environmental variable to ensure that any errors from the test are logged in
83 the dashboard (all of the ctest wrapper scripts set it). The '-R <pattern>'
84 option gives ctest a regular expression. Any tests with a name as defined in
85 DartTestfile.txt that matches the pattern will be run.
87 Note that you must have run ./scripts/do_sed at least once already in order to
88 use Experimental mode.
90 Updating and Building Within CTest
91 ==================================================================
93 Before each Nightly run, ctest will automatically update the BACULA_SOURCE
94 directory, and submit these updates along with the test results. Any
95 Experimental runs will not.
97 Before either type of run actually begins running tests, ctest will run the
98 script scripts/update-ctest. This script first compares the svn version of
99 BUILD_SOURCE with that of the build/ directory. If the two versions differ, or
100 if the build/ directory does not exist, it will automatically run 'make setup'
103 Viewing the Dashboard
104 ==================================================================
106 You can view the dashboard at:
108 http://regress.bacula.org:8081
110 Results will not be visible as soon as they are submitted to the server.
111 Processing is currently done every 10 minutes, so you may have to wait up to 15
112 minutes or so before your results show up.
114 =========================================================
116 Email from Frank describing the flow when running a ctest and some of the
117 problems that come up.
119 0. Start off with a local svn repository at version A, and the master
120 repository at version B.
122 1. nightly-disk is started.
124 2. nightly-disk runs scripts/config_dart
126 3. config_dart runs scripts/create_sed
128 4. create_sed pulls bversion and bdate out of the current local repo, so gets
131 5. config_dart then creates DartConfiguration.tcl from the .in file, leaving a
132 BuildName parameter of A.
134 6. nightly-disk then runs 'ctest -D Nightly'. This implicitly tells ctest to
135 perform Update, Configure, Build, Test, and Submit stages, in that order.
137 7. The Update stage runs 'svn update' on the local repository. The local
138 repository is now updated to version B from the master, but since the
139 DartConfiguration.tcl file was already created and has not been updated, the
140 Update.xml file has the version A BuildName still.
142 8. Next, the Configure stage runs. Since the configure process is handled in
143 tandem with the build process by 'make setup', this just calls /bin/true so as
144 to not throw any false errors, and can effectively be treated as a no-op.
146 9. Next is the Build stage, which is handled by calling scripts/update-ctest.
148 10. update-ctest checks the svn versions of regress/build vs BACULA_SOURCE.
149 Since the two are different (regress/build is still version A, while
150 BACULA_SOURCE has been updated to B) it calls 'make setup'.
152 11. 'make setup' copies BACULA_SOURCE to regress/build and configures and
153 builds it. It then calls scripts/do_sed
155 12. do_sed calls scripts/config_dart again. Since regress/build has been
156 updated to B, it regenerates DartConfiguration.tcl with a version B BuildName.
158 13. ctest now generates the Build.xml results file, but since BuildName was
159 still A while this stage began, this is what appears in the XML BuildName.
161 14. Done with the Build stage, ctest moves on to the Test stage, where it
162 actually calls the various test/ scripts as defined by DartTestfile.txt and
163 filtered by the -R option. At this point, since ctest is beginning a new
164 stage, it appears to re-read DartConfiguration.tcl (I believe this is intended
165 to allow ctest to bootstrap itself in a virgin cmake managed source code tree,
166 where the test configuration should be generated by cmake). The final
167 Test.xml file, therefore, contains a version B BuildName string, as opposed to
168 all previous steps, which have version A.
170 15. Finally, ctest flings the results at the dashboard. The dashboard does
171 ignores in what order or grouping XML files are submitted, and instead uses
172 the site/buildname tuple to distinguish them (at least for Nightly runs, I'm
173 not so sure about Experimental ones). In this case, instead of one complete
174 run, the dashboard sees two runs, one of which is Update through Make, and a
175 second one which is Test only.
177 For a sample of the problem, take a look at the fsweetser 2.3 sqlite3 Nightly
180 http://regress.bacula.org:8081/Bacula/Dashboard/Dashboard?trackid=29
182 The Update and Build information show up with a BuildName of
183 bacula-2.3.10-26Feb08-Linux-sqlite3, then after svn update hit the Test
184 information shows up with bacula-2.3.11-03Mar08-Linux-sqlite3. (Ignore for
185 the moment the fact my timestamps are at 6:59PM, rather than at 9PM where
186 they're supposed to be; this seems to be a Fedora specific client side issue I
187 haven't tracked down yet.)
189 Looking more closely at the test submitted to the public dashboard
190 (http://public.kitware.com/dashboard.php), I get the impression that the
191 BuildName parameter was misnamed, and intended to be treated more as a build
192 platform name, rather than the name of the build being tested. Rather than
193 create a hook to tag the version being tested, everyone as far as I can tell
194 just seems to rely on the timestamp of the test.
197 NOTE !!!!!!!!! ctest can actually back out changes that have been made to
198 your local source repository. As a consequence, it is probably better not to
199 use a directory in which you are developing code for Nightly tests. Seee the
200 below explanation given by Frank Sweetser.
202 When a Nightly run is done, the timestamp is set to the last occurring
203 instance of the time defined by the NightlyStartTime parameter. The piece
204 that I missed is that, in addition to using that timestamp for reporting to
205 the dashboard, the update stage also uses that point in time to determine
206 exactly which version of the repository to check out.
208 So if you make commit changes at 10PM EST, and then run a Nightly test run,
209 the NightlyStartTime of 9PM EST will back out those changes in the local
210 repository. Any subsequent runs that are started at 9PM EST the following day
211 or later will include them. This implies to me that NightlyStartTime should
212 be set such that you don't expect any developers to commit any changes in
213 between NightlyStartTime and the time at which the ctest run actually starts.
215 The alternative is to make use of the Experimental track. While it normally
216 just uses the local source tree as is, you can manually have it update:
218 ctest -D ExperimentalUpdate
220 Unlike Nightly, this will update to whatever the latest version of the