Hey Julian,
Thanks for the tip! Wercker clones with a depth of 50 by default. I have
now added a few more commands to convert the shallow clone to a full clone.
I am now able to run the tests and get them passing. Wercker is also
quite stable, I ran the tests a few times and they all pass each time.
Compare to travis, wercker is also much faster, tests take about 9
minutes to run compared to the 25ish minutes with travis.
Would you be interested in moving the CI over to wercker? You can check
out my branch here: https://github.com/F21/calcite/tree/use-wercker
(.travis.yml is replaced with wercker.yml).
Cheers,
Francis
On 5/10/2016 3:08 AM, Julian Hyde wrote:
Francis,
Try removing the depth argument when you do 'git clone'.
Julian
On Oct 3, 2016, at 21:30, F21 <[email protected]> wrote:
I tried to force travis to use a standard trusty vm in
https://github.com/apache/calcite/pull/293, so that the VM is assigned 7.5GB of
memory. Unfortunately, travis is still quite flaky, with tests only passing
some of the time.
Would it be useful to move to another CI? I tried setting up my fork to use
wercker: https://github.com/F21/calcite/blob/use-wercker/wercker.yml which
failed here:
https://app.wercker.com/F21/calcite/runs/build-jdk-7/57f3172e3b89220100273603?step=57f3173814732000019140d0
Unfortunately, I haven't found a solution to this error:
[ERROR] Failed to execute goal
pl.project13.maven:git-commit-id-plugin:2.1.9:revision (default) on project
calcite: Execution default of goal
pl.project13.maven:git-commit-id-plugin:2.1.9:revision failed: Unable to calculate
distance between [commit be18b25415a305aab2d0de2bd074755630e08462 1475548960
----sp] and [commit 08c56b158ffcfcf205a919cc9fff77a692e649f6 1474099831 ----sp]:
Missing commit 63c51d0c6459a4de5cab01188a7f3b7dd1a259fb -> [Help 1]
Francis
On 4/10/2016 10:55 AM, Julian Hyde wrote:
I’ve noticed it. A lot of Travis builds seem to hang. See the last two lines of
https://travis-ci.org/apache/calcite/jobs/163844678:
<https://travis-ci.org/apache/calcite/jobs/163844678:>
No output has been received in the last 10 minutes, this potentially
indicates a stalled build or something wrong with the build itself.
The build has been terminated
I don’t know why Travis does this. I suspect that the Travis VM doesn’t have
enough memory to complete the task, or the job takes longer than Travis allows,
but these are wild guesses.
I have been running the build and tests on my own Linux server in a variety of
configurations (in particular JDK 1.7 and 1.8, and a couple of timezones, and
occasionally also on macOS and Windows), and I can confirm that everything is
stable.
Julian
On Oct 3, 2016, at 3:20 PM, Laurent Goujon <[email protected]> wrote:
Hi,
I just opened a pull request against Calcite master, and the travis build
failed. But then I checked the master branch, and according to Travis, it's
also failing.
Is it something known, or just a transient issue?
Thanks in advance,
Laurent
