Closed umar456 closed 6 years ago
This part looks suspicious:
Using HTTP trigger method Trigger site: Trigger url: ?xmlfile=sigmaHAHA20181028-0524-ExperimentalXMLBuild.xml Error when triggering: ?xmlfile=sigmaHAHA20181028-0524-ExperimentalXMLBuild.xml
I presume you're using a CTest driver script that calls ctest_submit()
to upload results to CDash? If so, you probably need to modify the CTestConfig.cmake
file in your source directory.
Here's an example of what mine looks like when I'm submitting to my toy Docker instance:
set(CTEST_PROJECT_NAME "TestProject")
set(CTEST_NIGHTLY_START_TIME "01:00 UTC")
set(CTEST_DROP_METHOD "http")
set(CTEST_DROP_SITE "localhost:8080")
set(CTEST_DROP_LOCATION "/submit.php?project=TestProject")
set(CTEST_DROP_SITE_CDASH TRUE)
For this example I used ctest to upload the results. My main CTestConfig.cmake looks like this:
set(CTEST_PROJECT_NAME "ArrayFire")
set(CTEST_NIGHTLY_START_TIME "01:00:00 UTC")
set(CTEST_DROP_METHOD "http")
set(CTEST_DROP_SITE "67.207.87.39")
set(CTEST_DROP_LOCATION "/submit.php?project=ArrayFire")
set(CTEST_DROP_SITE_CDASH TRUE)
Of course I changed the drop site to localhost before I ran the ctest command. I am getting the issue using both the CTest scripts and the ctest command.
I just tried to submit another result to cdash on my local instance and I did not see the error on the client side but I still see the same issue in my original post. Here is the output of the ctest command:
ctest -VV -D Experimental -j24 -R basic_cpu
...
globbing for coverage in: /home/umar/devel/arrayfire/build3/test/gtest/googletest/CMakeFiles/rebuild_cache.dir
globbing for coverage in: /home/umar/devel/arrayfire/build3/test/gtest/googletest/CMakeFiles/test.dir
Cannot find any GCov coverage files.
Not a valid Intel Coverage command.
Cannot find any Python Trace.py coverage files.
Cannot find Cobertura XML file: /home/umar/devel/arrayfire/build3/coverage.xml
Cannot find GTM coverage file: /home/umar/devel/arrayfire/build3/gtm_coverage.mcov
Cannot find Cache coverage file: /home/umar/devel/arrayfire/build3/cache_coverage.cmcov
Cannot find Jacoco coverage files: /home/umar/devel/arrayfire/*jacoco.xml
Cannot find BlanketJS coverage files: /home/umar/devel/arrayfire/*.json
Cannot find Delphi coverage files: /home/umar/devel/arrayfire/build3/*(*.pas).html
Cannot find any coverage files. Ignoring Coverage request.
UpdateCTestConfiguration from :/home/umar/devel/arrayfire/build3/DartConfiguration.tcl
Parse Config file:/home/umar/devel/arrayfire/build3/DartConfiguration.tcl
Submit files (using http)
Using HTTP submit method
Drop site:http://localhost/submit.php?project=ArrayFire
Upload file: /home/umar/devel/arrayfire/build3/Testing/20181030-1704/Build.xml to http://localhost/submit.php?project=ArrayFire&FileName=sigma___HAHA___20181030-1704-Experimental___XML___Build.xml&MD5=f50918184e946174f82862b964798436 Size: 1036
Uploaded: /home/umar/devel/arrayfire/build3/Testing/20181030-1704/Build.xml
Upload file: /home/umar/devel/arrayfire/build3/Testing/20181030-1704/Configure.xml to http://localhost/submit.php?project=ArrayFire&FileName=sigma___HAHA___20181030-1704-Experimental___XML___Configure.xml&MD5=bc5e25ffbf974da8e21241688167f750 Size: 1470
Uploaded: /home/umar/devel/arrayfire/build3/Testing/20181030-1704/Configure.xml
Upload file: /home/umar/devel/arrayfire/build3/Testing/20181030-1704/Test.xml to http://localhost/submit.php?project=ArrayFire&FileName=sigma___HAHA___20181030-1704-Experimental___XML___Test.xml&MD5=ac53b2b02f855bd9d3d5306b72e58ec6 Size: 2255
Uploaded: /home/umar/devel/arrayfire/build3/Testing/20181030-1704/Test.xml
Submission successful
"GET /submit.php?project=ArrayFire&FileName=sigmaHAHA20181028-0524-ExperimentalXMLConfigure.xml&MD5=e2ce728f2cf4a7ce4273967a435f034c/ajax/dailyupdatescurl.php?projectid=1"
I see now. This is a bug with get_server_URI
for when $CDASH_BASE_URL
is not set. Instead of /ajax/dailyupdatescurl.php
we're accessing /submit.php
with a bunch of random junk in the query string. This ultimately results in another request back to submit.php
, which results in the full Apache logs you're experiencing.
Thanks for reporting this, I'll work on a fix shortly.
Setting the $CDASH_BASE_URL fixes the problem. Thank you for looking into this!
Fixed in #758
I have a docker container running the latest version of CDash master(0b369f51158ec05c27760f3ca62c2a4016be1019) to maintain my project.
I have noticed that the space on the server keeps running out after a week or so and the utilization has spiked for extended period time. The space runs out because of the excessive logs that CDash creates. Here is an example:
Notice how the logs are repeating. This does not stop until the system runs out of space. The only workaround is to destroy and restart the container. This seems to happen every time I submit a build to CDash. I have recreated this on a clean container on my local machine.
CTest also displays the following error:
I can send you additional information if necessary.