Set up LCM CDash tests for procyon, proxima, antares and spica machines #57
Comments
|
@lxmota , can you please modify your cronjob on algol to execute the CDash scripts I created there when you get the chance? I noticed that that machine has not been pushing results to the CDash site, I think b/c the job was never turned on. Thanks. |
|
@ikalash Yes, I still have to do this. I want to make it so it works for Trilinos as well. |
|
Ok. The scripts I wrote currently will build both Trilinos and Albany, then run Albany scripts. Your cronjob script should just have: cd /home/lcm/LCM/nightlyCDash The scripts assume there is already a Trilinos and Albany checked out here: /home/lcm/LCM . It's worth trying to clean up the scripts to load all the paths from your modules, rather than have them hard-coded in ctest_nightly.cmake.frag. |
Here is what I would like to accomplish with this, but have been unable to do so for lack of in-depth knowledge of CTest: The files ctest.cmake and Project.xml in the Albany/doc/LCM/dashboards directory are templates that similarly to the script build.sh are copied or symlinked to the top level LCM directory. Not sure that Project.xml is needed because it appears that the values set in it can also be set within a Cmake script like ctest.cmake. When build.sh is run in test mode as test.sh, the templates ctest.make and Project.xml are copied into the build directory, for example albany-build-serial-gcc-release. Then build.sh replaces actual values for the build into the placeholder lcm_* strings in the templates. The script build.sh then proceeds to call CTest which in turn runs the tests and submits the results to the dashboard.
|
I'm trying to write CMake versions of the LCM build scripts in this branch: |
|
@lxmota, can we close this issue? It looks like there are nightly tests for proxima and algol. My guess is this is sufficient since the other machines are very similar (same OS, TPLs, etc.). |
|
Yes! It went under the radar. |
|
No problem. I was trying to clean up / close old issues that have been resolved. Thanks for closing it. |
Note: this should soon be removed, together with the getNode calls. The Tpetra maps can be built without passing an object of type Node anyway. For more explanation, see issue #57 and PR 060 in trilinos.
Note: this should soon be removed, together with the getNode calls. The Tpetra maps can be built without passing an object of type Node anyway. For more explanation, see issue #57 and PR 060 in trilinos.
Note: this should soon be removed, together with the getNode calls. The Tpetra maps can be built without passing an object of type Node anyway. For more explanation, see issue #57 and PR 060 in trilinos.
I will do this, once @lxmota has a chance to clean up the algol CDash tests I created to use the modules on the various LCM machines.
The text was updated successfully, but these errors were encountered: