- Oct 02, 2024
-
-
Georges Racinet authored
This is an unexpected consequence of the general move to heptapod/heptapod of the Docker image: its name was used in Python tests and changed there too (service aliases), changing the expected result as well
-
- Aug 30, 2024
-
-
Georges Racinet authored
Development is no longer happening at Octobus, but at Cloudcrane
-
- Mar 28, 2024
-
-
Georges Racinet authored
-
Georges Racinet authored
-
- Mar 27, 2024
-
-
Georges Racinet authored
-
Georges Racinet authored
-
Georges Racinet authored
Done in a single commit, because it keeps the 100% coverage without further effort. The retry logic is implemented as a generic decorator (probably repeating the work of hundreds of developers worldwide, but it does not matter much for so simple things), hence we will be able to reuse it much more if needed
-
Georges Racinet authored
While this is in part redundant with the various other log lines, it has the advantage of being machine oriented (easy to leverage by monitoring systems) and systematic, giving us information about the load even if not stalled.
-
- Nov 23, 2023
-
-
Georges Racinet authored
-
Georges Racinet authored
-
Georges Racinet authored
-
- Nov 14, 2023
-
-
Georges Racinet authored
This is the time between the launch request (Git push in the case of Clever), and the witnessing of the first line in the job trace reported to the coordinator. To do this, we had to include the launch time in the message payload for `LAUNCH_SUCCESS`.
-
Georges Racinet authored
Up to now we had the custom message from the remote CC Git repo about deployment ending without context in the console output as stderr The URL logged should be the one without credentials.
-
Georges Racinet authored
-
- Sep 15, 2023
-
-
Georges Racinet authored
-
Georges Racinet authored
-
Georges Racinet authored
It is interesting to be able to debug resources in weird states, but by default we would not want them to linger. Hence we will now decommission by default. This can be changed with the new `paas_decommission_launch_failures` global option.
-
- Sep 14, 2023
-
-
Georges Racinet authored
-
Georges Racinet authored
If we cannot even query the trace log, or if we fail to report the timeout of a launching job, we still have to remove the job from the pending queue, which is usually capped at a low value. Pending queue saturation means the runner is completely stuck. It seems that the move to GitLab 15.11 in Heptapod 0.38 has the consequence of changing behaviour of the trace log in some cases (token nulled, 403 error without any hint), making this problem more frequent. Closes #51
-
- Feb 21, 2023
-
-
Georges Racinet authored
Closes #45
-
- Feb 06, 2023
-
-
Georges Racinet authored
the `| xargs rm` can lead to failure in the case only one executable had been built.
-
Georges Racinet authored
The separate script for that not longer exists. Also we need an option to skip the pwsh in the helper image, since we didn't build it.
-
- Feb 03, 2023
-
-
Georges Racinet authored
oops, made it to 0.6.0
-
Georges Racinet authored
This matches the images now pushed by CI/CD to registry.heptapod.net
-
- Feb 06, 2023
-
-
Georges Racinet authored
-
- Feb 01, 2023
-
-
Georges Racinet authored
-
Georges Racinet authored
-
- Jan 17, 2023
-
-
Georges Racinet authored
This is probably for informative purposes only
-
Georges Racinet authored
This will allow us to start merging upstream v13.11 into the heptapod branch.
-
- Jan 14, 2023
-
-
Georges Racinet authored
-
Georges Racinet authored
-
Georges Racinet authored
It might have been removed by external means, or even right before a hard shutdown.
-
Georges Racinet authored
Nice one!
-
Georges Racinet authored
In case a job has an already launched PAAS resource (not to be confused with the job being fully launched, we really need to come up with a change in terminology), before this change the log would have said "Launching job.." but that was just acknowledging the start of `launch_job` which itself may do nothing else than resuming the tracking of the job trace.
-
Georges Racinet authored
We simply weren't loading the `launched` attribute, resulting in failed Git pushes (already mitigated by parent changeset).
-
Georges Racinet authored
In theory, Heptapod PAAS Runner state tracking should avoid attempting to deploy again an application. In practice, we currently have a silly bug about that (making easy to test for real that this change works). The bug will be fixed in next commit, leaving hopefully only exceptional cases (hard shutdown).
-
Georges Racinet authored
Using the new format already in service in the debug signal. Also, pending jobs are currently especially interesting, since we just implemented #42 and #39, hence logging their details at the `INFO` level.
-
Georges Racinet authored
Now that the launcher threads are long (they wait for full job startub, see #42) and interruptible, waiting for them has become useless: all it was doing was to block shutdown until general timout, hence creating doubt about the consequences for ininterruptible operations (there weren't any, actually).
-
Georges Racinet authored
The contract of `repr()` for `JobHandle` is not really appropriate, as we don't want to log the token. On the other hand we really nead more information about the state, and this is used with `--debug-signal`
-
- Jan 13, 2023
-
-
Georges Racinet authored
-