I apologize that I mistakenly sent the email to dev@ozone.  It is
totally my fault.

Tsz-Wo

On Thu, Aug 6, 2026 at 10:44 PM Sumit Agrawal <[email protected]>
wrote:

> Hi Nicholas,
>
> Thanks for the feedback and seems this feedback is just based on *only last
> interaction* for HDDS-14040 which induced the issue and where we tried to
> fix the actual issue.
>
> *The interaction was not good due to fact in difference in expection
> between us for discussion (since the problem is complicated),*
> - I prefer POC code for experiments and getting quick test result, but
> you prefer exact implementation and test result. But I feel we get same
> output, and its based on different working style to achieve same result.
> - You expect more details written on temporary doc and full flow to take
> decision, but as I know the flow related to that area, I tend to skip for
> discussion as seems un-necessary. Some deep unrelated behavior, as not in
> finger tip, needs relook code.
>
> *Coming to fact of process being followed for the bug:*
> HDDS-14040:
> - This was reviewed and approved by you as expert review for ratis & ozone
> integration and merged only after that.
> - Test cases are added with due-deligence, covering scenarios and covering
> scenario impact as thought during that time.
>
> Fix: all test cases are added and merged only after approval.
>
> *Improvement for Ratis integration with ozone for DN flow:*
> - Already lot of discussion has happened to handle flaky nature of ratis
> and ozone integration. Still no concrete solution.
>
>
> *Coming to the concerns you have mentioned:*
>
>
> *- Why did the change get merged?  No tests added along with the changes.*
> I have never merged any PR without the approval and being reviewed. If any
> facts, please share.
> Showing *urgency* to fix problem is justifiable (still taken 2 monts to get
> fixed) and expected for delivery.
>
> - When we were discussing the issue, I found he actually neither understood
> the code nor the problem,
> - Sumit initially proposed some other changes without trying to
> completely understand
> the code first.
> *This is a communication gap* of your expectation and my delivery as *doc*.
> I may know the flow as required for the bug, but you have asked for
> complete flow in doc as you might be trying to understand and to take
> decision for solution. This does not provide any outcome of my
> understanding.
>
> - When we came up with a potential bug fix, Sumit wanted to merge it
> without adding new tests.
> - When Sumit agreed to add tests, the tests he proposed again were way too
> simple.
> IMO, have added all required test cases and scenario as suggested. Adding
> test cases are never being objected and neither in comments being ignored.
> I do appreciate your opinion about getting tested on multiple internal
> environment.
>
> Still if any sceanrio is missed for test case in apache ozone, and need
> some improvement, eager to fix and handle. Please share if anything
> remaining as not handled.
>
> So I think all comments are based on different expectation that happened
> during discussion, but *Apache* *Ozone process is never compromised and
> eager to take expert opinion as I lack Ratis depth code flow*. And always
> there is an improvement.
>
> @Nicholas: Apache Ozone contribution can never be denied based on internal
> discussion and differences in opinion about the proceeding. Please recheck
> your opinion over this.
>
>
> Regards
> Sumit
>
> On Thu, Aug 6, 2026 at 1:49 PM Tsz Wo Sze <[email protected]> wrote:
>
> > Hi Sammi and others,
> >
> > Thanks for bringing up this discussion!
> >
> > I worked with him closely in the same company previously.  One of the
> > issues caused job slowness, which was reported by some users.  Sumit
> tried
> > to fix it.  He committed some code changes.  Unfortunately, that
> particular
> > change could cause data loss, a much more severe problem.  The good thing
> > was that, although the change was merged, the problem had been discovered
> > before causing real data loss.
> >
> > Of course, all developers including you and me are susceptible to
> > introducing bugs.  However, the manner in which the bug was introduced,
> > along with Sumit’s subsequent handling of the issue, has given rise to
> > significant concern.
> >
> > - Why did the change get merged?  No tests added along with the changes.
> > Sumit did test it manually but the manual tests conducted were way too
> > simple.
> >
> > - When we were discussing the issue, I found he actually
> > neither understood the code nor the problem, but he dared to merge
> > the code.  Note that the code he changed was at the most important data
> > path in Ozone.
> >
> > - Sumit initially proposed some other changes without trying to
> completely
> > understand the code first.
> >
> > - When we came up with a potential bug fix, Sumit wanted to merge it
> > without adding new tests.  The reason was that adding new tests was not
> > easy.  It was understandable but unacceptable.
> >
> > - When Sumit agreed to add tests, the tests he proposed again were way
> too
> > simple.
> >
> > We spent a long time having many discussions.  He finally was convinced
> --
> > for that particular issue, we had to understand the problem first and
> also
> > had to test the change seriously.
> >
> > I would agree that Sumit is smart and very good at debugging, and he has
> > made a lot of contributions to Ozone.  However, he might be immature to
> > handle important issues.  Therefore, I would disagree with nominating him
> > as a PMC member.
> >
> > Regards,
> > Tsz-Wo
> >
> >
> >
> >
> > On Thu, Jul 30, 2026 at 12:14 AM Sammi Chen <[email protected]>
> wrote:
> >
> > > Dear PMCs,
> > >
> > > I would like to propose adding Sumit Agrawal[1] to the PMC.
> > >
> > >  Sumit
> > > - has been a committer on 2023-05-22, 3 years ago [2]
> > > - has contributed 156 PRs [3], 476 reviews [4]
> > >
> > > These are the highlights of his contributions
> > > - Fix broken Quota function with FSO bucket and Legacy bucket, and
> later
> > > quota for snapshot support. [5][6]
> > > - Improve HDDS block deletion efficiency. [7]
> > > - Handling disk issues, preventing disk becomes full, and handle
> > duplicate
> > > container replicas on a datanode[8]
> > > - Distributed tracing improvement, migrate from OpenTracing to
> > > OpenTelemetry[9]
> > > - Migrating Ozone Manager replication from post Ratis execution to Pre
> > > Ratis execution, design discussion, and work on the POC. [10][11]
> > >
> > > Besides above feature based contributions, Sumit also has a lot of
> > > production cluster issue investigation and fix experiences, and
> > contributed
> > > a lot of critical fixes to Ozone,
> > > including performance bottleneck resolve, data loss prevention, etc.
> > > I think he is a good candidate for Ozone PMC. He will bring his unique
> > > insight to the community. What do you think?
> > >
> > > Disclaimer: We share the same employer.
> > >
> > > - Sammi Chen
> > >
> > > [1] https://github.com/sumitagrawl
> > > [2] https://whimsy.apache.org/roster/committer/sumitagrawal
> > > [3]
> > >
> https://github.com/apache/ozone/issues?q=is%3Apr%20author%3Asumitagrawl
> > > [4]
> > >
> >
> https://github.com/apache/ozone/pulls?q=is%3Apr+reviewed-by%3Asumitagrawl+
> > > [5] https://issues.apache.org/jira/browse/HDDS-7479
> > > [6] https://issues.apache.org/jira/browse/HDDS-7598
> > > [7] https://issues.apache.org/jira/browse/HDDS-6548
> > > [8] https://issues.apache.org/jira/browse/HDDS-12564
> > > [9] https://issues.apache.org/jira/browse/HDDS-13679
> > > [10] https://issues.apache.org/jira/browse/HDDS-11897
> > > [11] https://issues.apache.org/jira/browse/HDDS-11900
> > >
> >
>
>
> --
> *Sumit Agrawal* | Senior Staff Engineer
> cloudera.com <https://www.cloudera.com>
> [image: Cloudera] <https://www.cloudera.com/>
> [image: Cloudera on Twitter] <https://twitter.com/cloudera> [image:
> Cloudera on Facebook] <https://www.facebook.com/cloudera> [image: Cloudera
> on LinkedIn] <https://www.linkedin.com/company/cloudera>
> ------------------------------
>

Reply via email to