search

ros-infrastructure / buildfarm

Build scripts and notes for catkin debian build pipeline.
6 stars 15 forks source link

jenkins_support.py: If configured with bad URL, bad error message #117

Closed mirzashah closed 11 years ago

mirzashah commented 11 years ago

If you have an invalid URL, the error message printed is not clear why

dirk-thomas commented 11 years ago

Can you please mention what command you invoked and what the output was.

tfoote commented 11 years ago

It was create_*_jobs.py and it just errored as a 404 with a backtrace.

On Sat, May 4, 2013 at 10:21 AM, Dirk Thomas notifications@github.comwrote:

Can you please mention what command you invoked and what the output was.

— Reply to this email directly or view it on GitHubhttps://github.com/ros-infrastructure/buildfarm/issues/117#issuecomment-17437867 .

dirk-thomas commented 11 years ago

Is that a duplicate of #118?

tfoote commented 11 years ago

No, this is for the url in server.yaml, that is for the cache file.

On Tue, May 7, 2013 at 2:00 AM, Dirk Thomas notifications@github.comwrote:

Is that a duplicate of #118https://github.com/ros-infrastructure/buildfarm/issues/118 ?

— Reply to this email directly or view it on GitHubhttps://github.com/ros-infrastructure/buildfarm/issues/117#issuecomment-17531083 .

tfoote commented 11 years ago

I guess the patch for #118 will at least let the user see the url failing. It could be caught more more cleanly to avoid the backtrace, but that's probably not necessary for this script.

dirk-thomas commented 11 years ago

If it is related to a wrong URL in the server.yaml file #118 would not affect that behavior. The jenkins url is processed by the jenkins module and would raise a socket.error.

And what cache file are you referring to?

tfoote commented 11 years ago

Reopening if this won't hit the same code path.

The other error is loading the rosdistro cache.

dirk-thomas commented 11 years ago

Since this is only guessing currently: @mirzashah please mention which exact command you ran, what your configuration files look like and copy'n'paste the full output of the tool.

mirzashah commented 11 years ago

Hey @dirk-thomas, I don't fully remember. Both issues that I posted were at the request of @tfoote while he was helping us setup our own buildfarm and did not want to forget [potential] bugs as he caught them. From what I recall, we ran this script:

./scripts/create_release_job groovy --fqdn http://wbr.willowgarage.com/repos --commit --wet-only

Where "wbr" is a machine on our local machine. It then barfed with a huge stack trace that had something to do with a bad URL, I believe we passed in a malformed URL in the form: "http://http://wbr.willowgarage.com/repos" where we had repeated the http incorrectly. After we fixed it, things worked ok.

Hope that helps!

tfoote commented 11 years ago

my memory had the malformed url was in the server.yaml not the fqdn which is why jenkins_support was the issue.