Since the order was changed, the first values are actually the older ones.
This commit inverse newer_build_stats and older_build_stats
values in order to always have the new keys. Before this commit the new
keys where not displayed. A future improvement may be to combine keys
from all builds.
This commit also proposes to give a 0 value if the key did not exist in
the older build. This means that new keys will appear with a big
difference. This is maybe not a good idea and needs some testing. A
better solution would be to search for the first apparition.
The page of PRs in "error" is currently kinda broken: it does not show
any feedback aside from the PR being in error which is not very
useful.
The intent was always to show an explanation, but when adding the page
I just deref'd `staging_id` which always fails though in two different
ways:
* when the PR can not be staged at all (because of a conflict) there
is no staging at all with a reason to show, so there should be
a fallback that the PR could not even be staged
* `staging_id` is a related field which deref's to the staging_ids
of the first *active* batch, except when a staging completes
(successfully or not) both staging and batch are disabled.
Plus the first batch will be the one for the first staging so if the
PR is retried and fails again the wrong reason may be displayed.
So update the section to show what we want: the reason of the
staging of the *last* batch attached to the PR.
NOTE: there's one failure mode remaining, namely if a staging fails
then on retry the PR conflicts with the new state of the
repository (so it can't be staged at all), the "reason" will
remain that of the staging. This could be mitigated by attaching
a "nonsense" batch on failure to stage (similar to the
forwardport stuff), that batch would have no staging, therefore
no staging reason, therefore fallback.
Closes#525
Draft was added in 82174ae66e but turns
out the v13 ORM is not able to create a required column (even when
given a default value), at least for booleans.
So create it by hand.
On staging failure, the 'bot would point to the first error or failure
status it found on the commit. This turns out not to be correct as
we (now) have various statuses which are optional, and may fail
without blocking stagings (either because they're solely informational
or because they're blocking & overridable on PRs).
Fix this so the 'bot points to the first *required* failure.
Fixes#517
On conflicts in multi-commit PRs developers sometimes get confused as
to what happened why.
If a conflict occurs and the source pull request had multiple
commits, list all the source commits and show which one broke.
Related to #505
* Remove the forwardport creating PRs in draft, that was mostly to
avoid codeowners triggering but we've removed the github one and
hand-rolled it, so not a concern anymore.
* Prevent merging `draft` PRs, the mergebot rejects approval on draft
PRs and insults people.
TBD (maybe): try to create *conflicting* forward-port PRs in draft so
it's clearer they need to be *fixed*? Issue of not being able to do
that on all private repositories remains so~~
Fixes#500
"Uniquifier" commits were introduced to ensure branches of a staging
on which nothing had been staged would still be rebuilt properly.
This means technically the branches on which something had been
staged never *needed* a uniquifier, strictly speaking. And those lead
to extra building, because once the actually staged PRs get pushed
from staging to their final destination it's an unknown commit to the
runbot, which needs to rebuild it instead of being able to just use
the staging it already has.
Thus only add the uniquifier where it *might* be necessary:
technically the runbot should not manage this use case much better,
however there are still issues like an ancillary build working with
the same branch tip (e.g. the "current master") and sending a failure
result which would fail the entire staging. The uniquifier guards
against this issue.
Also update rebase semantics to always update the *commit date* of the
rebased commits: this ensures the tip commit is always "recent" in the
case of a rebase-ff (which is common as that's what single-commit PRs
do), as the runbot may skip commits it considers "old".
Also update some of the utility methods around repos / commits to be
simpler, and avoid assuming the result is JSON-decodable (sometimes it
is not).
Also update the handling of commit statuses using postgres' ON
CONFLICT and jsonb support, hopefully this improves (or even fixes)
the serialization errors. Should be compatible with 9.5 onwards which
is *ancient* at this point.
Fixes#509
Although it's possible to find what PR a commit was part of with a bit
of `git log` magic (e.g. `--ancestry-path COMMIT.. --reverse`) it's
not the most convenient, and many people don't know about it, leading
them to various debatable decisions to try and mitigate the issue,
such as tagging every commit in a PR with the PR's identity, which
then leads github to spam the PR itself with pingbacks from its own
commits. Which is great.
Add this information to the commits when rebasing them (and *only*
when rebasing them), using a `Part-of:` pseudo-header.
Fixes#482
Proper attribution is important in general, but especially for
external contributors. Before this change, and the previous change
fixing authorship deduplication, it was rather easy for a "squashed"
conflict commit (attributed to the 'bot for lack of a really clean
option) to get merged by mistake.
This commit changes two things:
* The mergebot now refuses to stage commits without an email set, the
github API rejects those commits anyway so any integration mode
other than `merge` would fail, just with a very unclear error
* The forwardbot now creates commits with an empty author / committer
email when the pull request as a whole has multiple authors /
committers. This leverages the mergebot update.
Also clean up the staging process to provide richer error reporting
using bespoke exceptions instead of simple assertions. I'm not sure
we've ever encountered most of these errors so they're really sanity
checks but the old reporting would be... less than great.
Fixes#505
a45f7260fa had intended to use the
original authorship information for conflict commit even if there were
multiple commits, as long as there was only one author (/ committer)
for the entire sequence.
Sadly the deduplication was buggy as it took the *authorship date* in
account, which basically ensured commits would never be considered as
having the same authorship outside of tests (where it was possible for
commits to be created at the same second).
Related to #505
If a PR is closed on github and unknown by the mergebot, when fetched
it should be properly sync'd as "closed" in the backend, otherwise the
PR can get in a weird state and cause issues.
Also move the "I fetched the thing" comment before the actual creation
of the PR for workflow clarity, otherwise the reader has the
impression that the 'bot knew about the PR then fetched it anyway.
And improve savepoint management around the fetching: savepoints
should be released in all cases.
Closes#488.
If two PRs have the same label *in different projects entirely*, the
mergebot should not consider them to be linked, but it did as shown by
the warning message on odoo-dev/odoo#905 (two PRs created from the
same branch in different projects were seen as linked by the status
checker).
3b417b16a1 fixed the actual staging
selection, it's only the warning which did not properly segregate PRs.
Only group PRs which target the same branch (therefore are within the
same project).
Fixes#487
Though the forwardport posts regular reminders that an fw is outdated,
it can be easy to miss for the non-subject (and apparently the
subjects often just ignore the information entirely).
Add a few relevant links there:
* on PR pages, add a link to either the source or the
forward-ports (if applicable), as well as the merge date
* add a new page which lists all the PRs with outstanding
forwardports, as well as the forwardports in question
Fixes#474
Don't try to parse the response as JSON in the error case(s): if the
errors are bad enough github can return complete non-parseable
garbage.
Only access the "text" property (response body, decoded, but unparsed)
in error cases, only parse in the success case.
Also avoid reusing variables for completely different values, even if
they're of the same type, especially if they can overlap.
fixes#470
Initial thinking was to remove the check entirely and leave it to the
mergebot, but the lack of error reporting / forwarding means while
technically correct it would probably be somewhat difficult to grok.
Instead, improve the error reporting:
* add a dedicated message when trying to r+ via fwbot on a non-fw
PR (note: maybe the fwbot should not care? and just send it as-is
to the mergebot in that case?)
* clarify the ACL error
* post both message as the forwardbot rather than the mergebot
Also add a missing token note for the feedback from the forwardport
limit.
fix#469
If a PR is updated and has extent forward-ports, those forwardports
get updated automatically ("followup").
However there is an issue if the udpate causes a conflict in the
followup: the conflict gets silently pushed, and may fairly easily get
merged if it occurs in an area which the CI doesn't cover.
It's unclear what the policy really should be for this issue, and
there is no real way to *block* a pull request at the moment (save by
putting it in error at the mergebot level I guess?), so for now
clearly notify the user on both the modified PR and the followup,
with a comment on both.
We may want to revisit this eventually.
Fixes#467
There was already a check, but the way the check behaved
means *detached* PRs would not be prevented from setting their
forward-port, despite that not doing anything.
Fix it by checking if the current PR has a source, not a parent.
Fixes#465
Previously, a PR's status page would only show the linked / related
PRs when `open`.
Since the relations between PRs remains useful, also make this
information available during staging and after merging.
Fixes#463
* in the main dashboard, show the exact UTC timestamp (with a Z
marker) on hover, not just the relative delta
* in the branch details page, show the full timestamp, zoned, in
ISO-8601 format
* The repo would only be registered at the very end of the creation,
meaning an error *during* the repo creation (e.g. while uploading the
first blob or setting up webhooks) would leave the repository
undeleted. Register the repository as soon as we know it was
created, in order to correctly dispose of it afterwards.
* Migrate logging.warning call to warnings.warn on repository deletion
failure: pytest will print warnings during its reporting, not so for
log warnings (?)
Since the frontend_assets are loaded with `defer="defer"`,
the page sometimes fail with the message:
```
stats.js:212 Uncaught ReferenceError: Chart is not defined
at updateChart (stats.js:212)
at stats.js:158
at XMLHttpRequest.xhttp.onreadystatechange (stats.js:53)
```
This commit checks that Chart is available before tring to render the graph.
Thanks to @kebeclibre for the help.
When getting pull info, the alive state can be determined easily,
meaning that this field can join the "_compute_branch_infos" familly
Hook was catching some changes made on pr and was conditionnaly updating some fields
and triggering some other operations conditionaly depending on the action flag.
All of the information needed to update the pull info should always be present in the
payload body, meaning that all fields can be updated at once in case some hook was missed,
and additionnal operation can be triggered based on fields changes.
In some conditions, it appears that a containerized build can eat up
all memory of the container host. This leads to disturbance of other
builds as the kernel OOM killer enters the dance.
With this commit, the docker ability to limit memory usage of a
container is used. The OOM killer will choose its victim among the
container processes.
The containers memory limit has to be set in the runbot settings. If not
set, no memory limit is used.
runbot servers are running with a log-level debug in order to have usefull
debug information, but this also causes some noise comming from odoo.
This pr changes most debug to info.
- Add draft pr management to avoid to trigger code owner on draft pr.
- Add check on falsy config on trigger id (avoid crash, usefull to disable trigger)
- Add extra params on custom trigger to avoid to write specific config every time.
- Trigger a new batch automaticaly when updating target/draft
Some projects may use a totally different Dockerfile. In order to avoid
new branches of those projects to automatically build with the generic
default Dockerfile, this commit adds the possibility to configure a
Default Dockerfile on a project.
Since Odoo 14.0, the recommended Ubuntu LTS release is Focal, the
default Docker should be updated accordingly.
A custom template with bionic have to be manually created on runbot
instances that still build Odoo < 14.0.
* A small change is made in the templates logic that builds the
Dockerfile: a `runbot_pip` dict entry now exists in order to install the
python libs required by the runbot. On the other hand, `additional_pip`
should only be used to install optional python libs for Odoo.
* Upgrade Chrome version to 90.0.4430.93-1 as this one is currently in
use on our current runbot instance. Just keep in mind that the Odoo
screencast feature does not work anymore since Chrome 88.
* gsfont is added because of a bug [0] that affects python-reportlab in
Focal.
* pyCrypto package is removed. It was used in an Odoo addon that
disappeared in odoo/odoo@2738341c21
* dbfread, websocket-client are now installed as a deb package as they exists in Bionic and
Focal
* pdfminer.six is now removed because a deb package exists in Focal but
not in Bionic. It means that it has to be added in deb_packages_python
in Dockerfiles for odoo > 13.0 and in additional_pip for odoo <= 13.0
[0]: https://bugs.launchpad.net/ubuntu/+source/python-reportlab/+bug/1918107
The test in the original code will never fire because the value searched
for is not in the keys of the dictionary, but in one of the lists which
are in the values. Work around this by maintaining a reverse dictionary
module name -> commit and use this for the test.
When a dockerfile to_build field is False for any reason and a new
runbot is setup, it's easy to miss the point and builds that involves
this particular Dockerfile will fail on the new runbot.
With this commit, the Dockerfiles tree view is improved to easily spot
those kind of problems:
- to_build field is visible and the Falsy lines are in yellow (warning)
- the empty Dockerfile's are in red (danger)
- versions are now visible in the tree view too
When a build slot is hidden in the batch tile but is responsible of the
batch failure, the failure reason may not be obvious for the user.
With this commit, an hidden slot appears if the slot build is in
failure.
Sometimes a pr pull info can fail.
- Most of the time it is only temporary and it will be successfull on next try.
- In some rare case the pr will always fail (github inconsistency) The pr exist in git but not on github api.
For this rare case, we store the pr in memory in order to unstuck other pr/branches update.
We consider that this error should not remain, in this case github needs to fix the inconsistency.
This is why the runbot model don't handle such a case for now.
Another solution would be to create the pr with fake pull info. This idea is not the best one
since we want to avoid to have many pr with fake pull_info in case of temporary failure of giothub services.
With this solution, the pr will be retried once every cron loop.
We dont except to have pr with this kind of persistent failure more than every few mounths/years.
