Suggestions needed

[Jeffrey Harris]

I'm trying to figure out a way to automate submitting a test to our hardware testbed and waiting for the test to complete so that the Jenkins job can report its status.

Are either of the ideas below possible?

Create a Jenkins job that submits the test (this is the easy part), waits until the job is informed the test is complete and then processes the results.

The other idea I've had is to create a spool_test job that only spoils the test. When the test is complete it starts a complete_test job to process the results.

Ultimately I'd like to have a regress_tests job that spools several tests. But I'm not sure with the second approach how to tell regress_tests that it's not done until all the complete_test jobs (that it doesn't directly start) are done.

Thanks for any help,
Jeff

[John Mellor]

How long does the hardware testing take? If its only a couple of hours, why not just run a shell step to ssh or whatever out to the hardware and test? A shell step normally blocks until completed, so the job will be in-progress until the tests are done.

--
You received this message because you are subscribed to the Google Groups "Jenkins Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to jenkinsci-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/jenkinsci-users/c7619784-6c20-471f-b727-de7bc57495ca%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Teichner Peter]

Just to add to that - make sure you set the timeout on the shell / ssh to unlimited otherwise it will time out

To view this discussion on the web visit https://groups.google.com/d/msgid/jenkinsci-users/5aaafbbef9dd47cca4abc2d90886539f%40mbx02cmb01p.esentire.local.

[Jeffrey Harris]

I think our longest test is 20 hours. Also, any number of tests can be spooled at a particular time.

The legacy system just puts the job in the queue and returns.

Jeff

[Logan MAUZAIZE]

I don't think blocking slave/agent for long duration is a good idea, except if a single test can be run at once for a particular platform. Thus, you must consider the hosting platform as a jenkins slave/agent and use a waiter script.

If you need feed back managed by Jenkins, may be you can use an additional job triggered using a project authentification token.