... | ... | @@ -88,7 +88,8 @@ In OpenMolcas there are two kinds of pipelines: |
|
|
|
|
|
Build pipelines are run every time you **push**, and their result is only
|
|
|
informative, whether they fail or pass does not have any effect. Verification
|
|
|
pipelines have to be started manually, e.g., through the web interface. One of
|
|
|
pipelines are run when you create a **merge request**, or push to an open
|
|
|
merge request; they can also be started manually, e.g., through the web interface. One of
|
|
|
the things MargePT2 does when handling merge requests is making sure a
|
|
|
verification pipeline is run. The result of the pipeline in this case
|
|
|
determines whether or not the merge request can be merged, this guarantees that
|
... | ... | @@ -112,7 +113,7 @@ In any kind of pipeline there can be failed jobs. Clicking on the job's button |
|
|
allows you to see a log and inspect what went wrong. You can also get an overview
|
|
|
of a pipeline's results by visiting [this page](https://molcas.altervista.org/tests/)
|
|
|
(note that it needs to pull the information from GitLab, so it is rather slow). It is
|
|
|
unfortunately not uncommon to have failures that are not related to the code changes
|
|
|
possible to have failures that are not related to the code changes
|
|
|
one has introduced. Some examples of this kind of error message are:
|
|
|
|
|
|
```
|
... | ... | @@ -126,10 +127,8 @@ ERROR: Job failed (system failure): error during connect: Get https://10.142.0.1 |
|
|
If this happens to one of your pipelines, you can and should rerun the job by
|
|
|
clicking the retry button. When possible, retry only the failing job, not the
|
|
|
full pipeline! Note that the `test:*` jobs require some files generated by the
|
|
|
corresponding `build:*` job, so if a `build:*` job failed, the `test:*` job
|
|
|
will probably fail too. Note also that the files generated by the `build:*`
|
|
|
jobs are only available for one day, so if they are needed after that the job
|
|
|
must be rerun.
|
|
|
corresponding `build:*` job, so in same cases it may be necessary to rerun
|
|
|
earlier jobs.
|
|
|
|
|
|
Another possible cause of failure is a timeout. Most jobs have a timeout of 4
|
|
|
hours, which should be enough, but sometimes a machine is very slow and does
|
... | ... | |