Testing the installation

It is very important that you verify that your DIRAC installation correctly reproduces the reference test set before running any production calculations. But also with all tests passing you are strongly advised to always carefully verify your results since we cannot guarantee a complete test coverage of all possible keyword combinations.

The test set driver is CTest which can be invoked with “make test” after building the code.

Environment variables for testing

Before testing with “make test” you should export the following environment variables:

$ export DIRAC_TMPDIR=/scratch            # scratch space (adapt the path of course)
$ export DIRAC_MPI_COMMAND="mpirun -np 8" # only relevant if you compile with MPI

Note that if you set the DIRAC_MPI_COMMAND the pam script will assume you that this is a parallel calculation. Before you test an MPI build, you have to set DIRAC_MPI_COMMAND otherwise the tests will be executed without MPI environment and many of the tests will fail.

Running the test set

You can run the whole test set either using:

$ make test

or directly through CTest:

$ ctest

Both are equivalent (“make test” runs CTest) but running CTest directly makes it easier to run sequential tests on several cores:

$ ctest -j4        # only for sequential tests

You can select the subset of tests by matching test names to a regular expression:

$ ctest -R dft

or matching a label:

$ ctest -L quick

To print all labels:

$ ctest --print-labels

How to get information about failing tests

We do our best to make sure that the release tarball is well tested on our machines. However, it is impossible for us to make sure that all functionalities and tests will work on all systems and all compiler versions. Therefore, some tests may fail and this is how it can look:

Total Test time (real) = 258.17 sec

The following tests FAILED:
         34 - dft_response (Failed)
         73 - xyz_input (Failed)
         75 - dft_ac (Failed)
Errors while running CTest

The first place to look for the reasons is build/Testing/Temporary:

$ ls -l Testing/Temporary/
total 108
-rw-rw-r-- 1 bast bast  2247 Dec 11 10:19 CTestCostData.txt
-rw-rw-r-- 1 bast bast 94798 Dec 11 10:19 LastTest.log
-rw-rw-r-- 1 bast bast    39 Dec 11 10:19 LastTestsFailed.log

LastTest.log contains a log of all tests that were run. Search for the test name or “failed”. If tests do not produce any output, then this is the place to look for reasons (environment variables, problems with pam).

In addition, for each test (whether passed or failed), a directory is created under build/test. So in this case you can go into build/test/dft_response to figure out what went wrong. The interesting files are the output files:

-rw-rw-r-- 1 bast bast 89538 Dec 11 10:17 lda_hf.out
-rw-rw-r-- 1 bast bast  1570 Dec 11 10:17 lda_hf.out.diff
-rw-rw-r-- 1 bast bast   841 Dec 11 10:17 lda_hf.out.reference
-rw-rw-r-- 1 bast bast   841 Dec 11 10:17 lda_hf.out.result

The *.out file is the output file of the test. The *.result file contains numbers extracted by the test script. The *.reference file contains numbers extracted from the reference file. The *.diff file is empty for a test where result and reference match. If they do not match, the *.diff file will list the lines of output where the results differ.