Failing ctest when using travis Trustry environment

AFNI Message Board
1

Greetings,

I just noticed that ctest shows failures on the travis ci environment (see https://travis-ci.org/afni/afni/builds/264515861):

23.28s$ ctest
Test project /home/travis/build/afni/afni/testing
Start 1: afni:start
1/2 Test #1: afni:start … Passed 5.98 sec
Start 2: do.examples
2/2 Test #2: do.examples …***Failed 17.24 sec
50% tests passed, 1 tests failed out of 2
Label Time Summary:
GLX = 17.24 sec
Total Test time (real) = 23.23 sec
The following tests FAILED:
2 - do.examples (Failed)
Errors while running CTest
The command “ctest” exited with 8.

As for the underlying cause, I noticed that the issue seems to have started when travis changed from the older “Precise” to the newer “Trustry” environment. A temporary fix would thus be to explicitly set the enivoronment to be “Precise” as explained here: https://blog.travis-ci.com/2017-07-11-trusty-as-default-linux-is-coming. This would require just adding a single line to .travis.yml, and then travis is again useful to catch compilation errors (amongst other things).

For the long term it would be preferably to find out why the second do.examples fails and fix the underlying issue - that way the tests also work using Trusty, and use of travis to test AFNI is more “future-proof”.

2

Hi Nick,

Thanks for the details. But note that we are not actually maintaining
those travis tests. I have no idea what the do.examples script is
running…

  • rick
3

rick reynolds Wrote:

Thanks for the details. But note that we are not
actually maintaining
those travis tests. I have no idea what the
do.examples script is
running…

All right, I re-raised the issue on github (https://github.com/afni/afni/issues/56) and ping-ed Yarik - hopefully with his help we can find out why the test fails and fix the issue.

4

Thanks to verbose output enabled by Yarik, the error reported is:

XIO: fatal IO error 11 (Resource temporarily unavailable) on X server “:20”
2: after 7415 requests (7398 known processed) with 0 events remaining.

See https://github.com/afni/afni/pull/57 for more details.

Could this be related to afni’s new detach-from-terminal-by-default behaviour that may cause a similar error (https://afni.nimh.nih.gov/afni/community/board/read.php?1,146837,146837)? Other ideas what may cause this error?

5

The error looks like it really happens just before with the count command, This looks like another issue posted today:

https://afni.nimh.nih.gov/afni/community/board/read.php?1,155796,155814#msg-155814

See “which count” and “where count” and make sure afni’s count is the one used in the script if that’s what you want.

6

Daniel Glen Wrote:

The error looks like it really happens just before
with the count command,

Thanks Daniel, indeed it seems a PATH issue.
Please see https://github.com/afni/afni/pull/58 for a Pull Request that should address the issue.
If you think it does, could you please merge?

7

nick Wrote:

Daniel Glen Wrote:

The error looks like it really happens just
before
with the count command,

Thanks Daniel, indeed it seems a PATH issue.
Please see https://github.com/afni/afni/pull/58
for a Pull Request that should address the issue.
If you think it does, could you please merge?

Thanks for merging - the PR did indeed resolve the issue.

For reference, the fix was changing the PATH variable so that the afni binary directory comes /first/, which probably deals with another conflicting ‘count’ command on the newer Ubuntu “Trusty” distributions. Earlier Ubuntu “Precise” did not seem to have a conflicting ‘count’ command.

For a related report and identical fix, see also https://afni.nimh.nih.gov/afni/community/board/read.php?1,155796,155796#msg-155796

8

Thanks for the update, Nick!

  • rick