[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [DISCUSS] Reverting commits on green post-commit status

Following the discussion, we are discussing as how to address failures 
on external projects caused by changes to Beam project.

I believe that that rollback first in case of red pre/postcommit tests is a
valid option, since it blocks Beam development process.

In case if downstream project has test failure outside of Beam project,
doing a rollback in Beam can be really frustrating for Beam developers.

I'd support Ken and Robert on this:
1. It is best to start dev-list discussion for visibility and for a decision on 
further course of action.
2. Interested party is motivated expose or implement relevant testm(s) 
inside Beam testing suites to prevent similar issue happening again. 
This will also simplify the rollback question since the policy will be 
applied due to tests going red. Ideally adding test will go together with 
starting the discussion on dev-list and request of rollback.


Have feedback

On Tue, Nov 20, 2018 at 1:15 PM Ruoyun Huang <ruoyun@xxxxxxxxxx> wrote:
The instructions in the post-commit policies page [1] is helpful, by not clear enough regarding what 'Rollback First' exactly means in the case of breaking a downstream project.  The discussions in this thread makes things more well defined.  I summarized things as an updating PR [2], please suggest if I missed anything, or whether we need that update at all.   

Also, one idea is we can have a time limit finding alternatives instead of rollbacks. That makes the steps more explicit and actionable, thus less likely anyone gets bad feelings, which then partially solves the concern that Maximilian brought up. Something like:  (4) If coulnd't figure out root cause within 2 hours, a rollback automatically becomes the best option.   Though that also depends on how much urgency such situation requires. 

On Mon, Nov 19, 2018 at 1:49 PM Maximilian Michels <mxm@xxxxxxxxxx> wrote:
The way I read Thomas' original email is that it's generally not a nice
sign for a contributor if her work gets reverted. We all come from
different backgrounds. For some, reverting is just a tool to get the job
done, for others it might come across as offensive.

I know of communities where reverting is the absolute last resort. Now,
Beam needs to find its own way. I think there are definitive advantages
to reverting quickly.

In the most obvious case, when our tests are broken and a fix is not
viable, reverting unblocks other contributors to test their code. I
think this has been working fine in the past.

In the less obvious case, an external Runner or system is broken due to
an update in the master. IMHO this does not warrant an immediate revert
on its own. As already mentioned, there should be some justification for
a rollback. This is not to make people's life harder but to figure out
whether the problem can be solved upstream or downstream, or with a
combination of both.

I think Thomas wanted to address this latter case. It seems like we're
all more or less on the same page. The core problem is more related to
communicating reverts in a way that helps contributors to save face and
the community to work efficiently.


On 19.11.18 10:51, Robert Bradshaw wrote:
> If something breaks Beam's post (or especially pre) commit tests, I
> agree that rollback is typically the best option and can be done
> quickly. The situation is totally different if it breaks downstream
> projects in which Kenn's three points are good criteria for determining
> if we should rollback, which should not be assumed to be the default option.
> I would say the root cause of the problem is insufficient visibility and
> testing. If external-to-beam tests (or production jobs) are broken in
> such a way that rollback is desired, I would say the onus (maybe not a
> hard requirement, but a high bar for exceptions) is on whoever is asking
> for the rollback to create and submit an external test that demonstrates
> the issue. It is their choice whether this is easier than rolling
> forward or otherwise working around the breakage. This seems like the
> only long-term sustainable option and should get us out of this bad
> situation.
> (As an aside, the bar for rolling back a runner-specific PR that brake
> that runner may be lower, though still not automatic as other changes
> may depend on it.)
> - Robert
> On Sat, Nov 17, 2018 at 7:35 PM Kenneth Knowles <kenn@xxxxxxxxxx
> <mailto:kenn@xxxxxxxxxx>> wrote:
>     Just adapting my PR commentary to this thread:
>     Our rollback first policy cannot apply to a change that passes all
>     of Beam's postcommit tests. It *does* apply to Beam's postcommit
>     suites for each and every runner; they are equally important in this
>     regard.
>     The purpose of rapid rollback without discussion is foremost to
>     restore test signal and not to disrupt the work of other
>     contributors, that is why it is OK to roll back before figuring out
>     if the change was actually bad. If that isn't at stake, the policy
>     doesn't make sense to apply.
>     But...
>       - We have at least three examples of runners where there are
>     probably tests outside the Beam repo: Dataflow, Samza runner, and
>     IBM Streams.
>       - We also may have users that try running their production loads
>     against Beam master branch to learn early whether the next release
>     will break them.
>     These are success stories for Beam. We should respect these other
>     sources of information for what they are: users and vendors giving
>     us a heads up about changes that will be a problem if we release them.
>     Often rollback is still a good option but IMO it is no longer
>     automatically the best option, and may not even be OK. I would say
>     that the case must be made clearly and *publicly* that
>     (1) something is actually broken
>     (2) the revert fixes the problem
>     (3) revert is the best option
>     In this scenario there is time to consider. An important and common
>     case is that a perfectly fine change exposes something already
>     broken, so the best option may be sickbaying downstream or pinning
>     their version/commit of Beam until they can fix.
>     Kenn
>     On Fri, Nov 16, 2018 at 8:15 PM Ahmet Altay <altay@xxxxxxxxxx
>     <mailto:altay@xxxxxxxxxx>> wrote:
>      >
>      > It sounds like we are in agreement that addressing issues sooner
>     is better. I think reverting is in general the less stressful option
>     because it allows a solution to be developed in parallel. Even with
>     that, it is not the only option we have and based on the severity
>     and the complexity of the problem we can consider other options.
>     Fixing forward might be feasible in some cases.
>      >
>      > We can bring issues back to the mailing list. This would be akin
>     to bringing any issues to the mailing list. I think JIRA is a better
>     tool for that. These reverts are happening because of an issue, and
>     JIRA allows informing all involved parties, creates emails to the
>     issues list for later searching through mailing archives, and
>     creates a record of things in structured way with components.
>      >
>      > We could establish general policies about for all reverts to have
>     an issue (which we already do because they are regular PRs),
>     including all people in the discussion (including the author and
>     reviewers) and follow up with new tests to expands Beam's test coverage.
>      >
>      > On Fri, Nov 16, 2018 at 7:55 PM, Thomas Weise
>     <thomas.weise@xxxxxxxxx <mailto:thomas.weise@xxxxxxxxx>> wrote:
>      >>
>      >>
>      >>
>      >> On Fri, Nov 16, 2018 at 7:39 PM Ahmet Altay <altay@xxxxxxxxxx
>     <mailto:altay@xxxxxxxxxx>> wrote:
>      >>>
>      >>> Thank you for bringing this discussion back to the mailing list.
>      >>>
>      >>> On Fri, Nov 16, 2018 at 6:49 PM, Thomas Weise <thw@xxxxxxxxxx
>     <mailto:thw@xxxxxxxxxx>> wrote:
>      >>>>
>      >>>> We have observed instances of changes being reverted in master
>     that have been authored following the contributor guidelines and
>     pass all tests (post commit). While we generally seem to have quite
>     a bit of revert action happening [1], this thread is about those
>     instances that are outside of our documented policies.
>      >>>>
>      >>>> For a contributor, it isn't a good experience to see reverts
>     (especially not out of the blue) after a PR has been reviewed, all
>     tests pass and generally care has been taken to do the right things.
>      >>>
>      >>>
>      >>> I completely agree. Everyone involved needs to have the context
>     about why a change is being reverted. A JIRA with information is
>     probably a good way to do it, similar to the any other issue we track.
>      >>>
>      >>>>
>      >>>>
>      >>>> Changes can have unforeseen effects downstream. Usually
>     releases provide the opportunity to mitigate such issues, not
>     necessarily by a revert, but in many cases by another change that
>     keeps everyone happy. Unexpected reverts can break someone else and
>     are disruptive.
>      >>>
>      >>>
>      >>> I disagree about waiting until release to resolve an identified
>     issue. Let's say we become aware of an issue through channels other
>     than Beam tests (e.g. user reports, a contributor running into
>     something not captured in Beam tests) and we know that it is a
>     credible issue that will block the release anyway. Addressing the
>     issue sooner will be less painful than addressing later. (I am not
>     suggesting addressing every such issue similar to we do not block
>     releases on every open issue. There needs to be a due diligence to
>     understand the severity and the impact.)
>      >>>
>      >>> We can improve on the above process. If we end up reverting a
>     change as a result of a report that is not covered by existing Beam
>     tests, we could expand the tests to catch same class of problems
>     even earlier by means of Beam's own tests.
>      >>
>      >>
>      >> I'm referring to the release process as an example how such
>     issues can be addressed. I'm not suggesting to wait until a release
>     to address issues; as you say the sooner they are identified the better.
>      >>
>      >> However, I don't agree with taking out the revert hammer on the
>     slightest sign that there could be a downstream problem. There are
>     better ways to handle this. First of all, I think that these issues
>     should be visible on the dev mailing list so that there is more
>     awareness overall (which will lead to better test coverage and
>     useful feedback in general). Second, we should make an effort to
>     resolve issues in a forward looking way. If after all it turns out
>     that a revert is most appropriate for the situation, it should
>     follow explicit agreement.
>      >>
>      >>>
>      >>>
>      >>>>
>      >>>>
>      >>>> Some discussion already took place on one specific PR [3], but
>     that is just an example and by no means an isolated incident.
>      >>>>
>      >>>> On some of these reverts, "internal" problems
>      >>>
>      >>>
>      >>> This is a communication issue that needs be addressed. Over
>     time we are getting new contributors and that is great, but it also
>     means these new folks need to understand operating conditions of
>     this community. Giving direct feedback would generally be most
>     efficient here.
>      >>>
>      >>>>
>      >>>> with an important runner are cited, with little to no
>     explanation. It would be nice if folks with more insight can shed
>     more light on this.
>      >>>
>      >>>
>      >>> All runners we accepted at out master branch and include in our
>     releases is important. I do not find a reference to an important
>     runner in the examples other than a question about what would the
>     outcome be for a runner other than Dataflow. I think the outcome
>     would be the same, and in my opinion it should be. We need to be
>     careful not breaking any runner we support.
>      >>>
>      >>>>
>      >>>> I hope that as outcome of this discussion, we can arrive at a
>     better understanding of why and when such reverts were necessary and
>     possibly find ways to largely avoid them going forward.
>      >>>>
>      >>>> Thanks!
>      >>>> Thomas
>      >>>>
>      >>>>
>      >>>> [1]
>     https://github.com/apache/beam/search?o=desc&q=Revert&s=committer-date&type=Commits
>      >>>> [2] https://beam.apache.org/contribute/postcommits-policies/
>      >>>> [3] https://github.com/apache/beam/pull/7012
>      >>>>
>      >>>

Ruoyun  Huang