[rdo-list] Using tripleo-quickstart to debug failures in a deployment

[Wesley Hayutin]

Greetings,

I thought it might be worth starting a discussion regarding including debug
steps as part of oooq. To define what I mean by a debug step I'm referring
to steps like [1], when something fails present more information about the
failure in the logs or console.

There are three primary use cases for oooq atm.

Use Cases:

1. new user deploying TripleO
2. CI
3. Developer debug/test

For use cases 2,3 extra debug information is very handy.  My own experience
with debugging 10-30 jobs a day extra debug information in the logs can
save a lot of time and prevent a misdiagnosis of a failed deployment.

For use case #1, I think we want to be careful and protect the simplicity
of the jinja templated bash scripts.  Basically presenting the user with a
clean and simple workflow.  I do think if a new user does unfortunately hit
an error, they might be completely unaware of how to debug the issue.  The
implementation here is up to the technical guys of course, but my thought
was to only run debugging tasks on failed tasks.

So I would classify additional debug steps as *critical* for #2, and a
really nice to have for #1, and #3.

I hoping that the group finds debugging steps as useful as I do.  I am also
hoping that we can review, add the following reviews [2-3].  I hope that
other people have additional debug steps to add, and we can start moving
debug tribal knowledge into the steps in tripleo-quickstart.

Thanks!

[1]
https://github.com/openstack/tripleo-quickstart/blob/master/roles/tripleo/overcloud/templates/overcloud-deploy.sh.j2#L56
[2] https://review.openstack.org/#/c/345559/
[3] https://review.openstack.org/#/c/346889/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rdoproject.org/pipermail/dev/attachments/20160727/949fd16b/attachment.html>