Red Hat Bugzilla – Bug 603719
Providing parameters to a test has no effect
Last modified: 2016-05-31 21:39:49 EDT
Description of problem:
I've scheduled a job with a parameter passed to the test. However,
it seems it did not have any effect.
Version-Release number of selected component (if applicable):
Version - 0.5.42
Snip from the job XML:
<task name="/CoreOS/httpd/Multihost/stress" role="SERVERS">
<param name="PROTOCOLS" value="IPv4 IPv6"/>
The actual job run IPv4 test only. Manual testing with
export TEST_PARAM_PROTOCOLS="IPv4 IPv6"
works as expected and runs both IPv4 and IPv6 subtests.
Passing parameters confirmed to work in RHTS. Marking as
regression and suggesting a blocker.
Can you tell me how you ran this job? using the new bkr workflow-simple command? I think I may have not implemented --task-param correctly. At least not like old single_workflow.
If you notice above the param is PROTOCOLS in the xml and your test seems to expect TEST_PARAM_PROTOCOLS. Should I be pre-pending TEST_PARAM_ to? Seems bad to me to hard code that.
I've cloned the job and pre-pended TEST_PARAM_ to see if that will work.
I confirmed that it works when the correct param is passed.
I'm just waiting on feedback on how this should be dealt with. I don't think pre-pending TEST_PARAM_ to everything passed in is a great idea. What if I have a test that expects a param that doesn't have TEST_PARAM_ as the begining?
Oh, I see. So it's only a different behavior in Beaker. But I
must admit I like the new way better. +1 for not prepending the
awkward prefix. Perhaps, for backward compatibility the
TEST_PARAM_* should be set as well. But I don't know how many
people/old tests this affects. I can happily live without the
ugly prefix :-)
By the way, this should be documented. I could not find any
mention about passing paramaters to tests in the User Guide.
Can you document?
+1 on documenting this. I just pinged around looking for this info and found it here or all places :)