You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@kafka.apache.org by Anna Povzner <an...@confluent.io> on 2018/04/14 22:54:06 UTC
[VOTE] KIP-279: Fix log divergence between leader and follower after
fast leader fail over
Hi All,
I would like to start the vote on KIP-279: Fix log divergence between
leader and follower after fast leader fail over.
For reference, here's the KIP wiki:
https://cwiki.apache.org/confluence/display/KAFKA/KIP-279%3A+Fix+log+divergence+between+leader+and+follower+after+fast+leader+fail+over
and discussion thread:
https://www.mail-archive.com/dev@kafka.apache.org/msg86753.html
Thanks,
Anna
Re: [VOTE] KIP-279: Fix log divergence between leader and follower
after fast leader fail over
Posted by Anna Povzner <an...@confluent.io>.
The vote for KIP-279 has passed with 5 binding and 2 non-binding +1s (and
no objections).
Thanks everyone for your reviews and feedback,
Anna
On Tue, Apr 17, 2018 at 1:49 PM, Anna Povzner <an...@confluent.io> wrote:
> Guozhang, thanks for catching this, I fixed the description (the example
> assumed response with 21, '11' was a typo).
>
> On Tue, Apr 17, 2018 at 1:25 PM, Anna Povzner <an...@confluent.io> wrote:
>
>> Hi Colin,
>>
>> Yes, the impact of "losing" entries in the LeaderEpoch file is more
>> round-trips for OffsetForLeaderEpoch. There is JIRA for the log cleaner:
>> https://issues.apache.org/jira/browse/KAFKA-6780, and to investigate if
>> there is actual possibility of losing committed records due to cleaning
>> further than high watermark. In any case, this KIP does not make it any
>> more or less likely.
>>
>> Thanks,
>> Anna
>>
>>
>>
>> On Tue, Apr 17, 2018 at 11:53 AM, Colin McCabe <cm...@apache.org>
>> wrote:
>>
>>> Thanks, Anna, this looks great.
>>>
>>> From the KIP:
>>>
>>> > Impact of topic compaction
>>> >
>>> > The proposed solution requires that we preserve history in
>>> > LeaderEpochSequence file. Note that this is also required in the
>>> current
>>> > implementation if we want to guarantee no log divergence. The only
>>> reason
>>> > for "losing" entries in LeaderEpoch file is if we actually lose
>>> > LeaderEpoch file and have to rebuild it from the log. If we delete
>>> all
>>> > offsets for a particular epoch for some topic partition, we may miss
>>> some
>>> > entries in the LeaderEpochSequence file.
>>> >
>>> > We will not do any changes to compaction logic in this KIP, but here
>>> is
>>> > possible fixes to compaction logic:
>>> >
>>> > Leave a tombstone in the log if we delete all offsets for some
>>> epoch,
>>> > so that LeaderEpoch file can be rebuilt
>>> > Do not compact further than persistent HW.
>>>
>>> Sorry if this has been answered before (I didn't find it in the DISCUSS
>>> thread) but what is the impact of "losing" these entries in the LeaderEpoch
>>> file? I suppose it would mean that more round-trips for
>>> OffsetForLeaderEpoch might be required during a leader change. That's the
>>> only impact, right?
>>>
>>> best,
>>> Colin
>>>
>>> On Mon, Apr 16, 2018, at 09:48, Ismael Juma wrote:
>>> > Thanks for the detailed KIP. +1 (binding)
>>> >
>>> > Ismael
>>> >
>>> > On Sat, Apr 14, 2018 at 3:54 PM, Anna Povzner <an...@confluent.io>
>>> wrote:
>>> >
>>> > > Hi All,
>>> > >
>>> > >
>>> > > I would like to start the vote on KIP-279: Fix log divergence between
>>> > > leader and follower after fast leader fail over.
>>> > >
>>> > >
>>> > > For reference, here's the KIP wiki:
>>> > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
>>> > > 279%3A+Fix+log+divergence+between+leader+and+follower+
>>> > > after+fast+leader+fail+over
>>> > >
>>> > >
>>> > >
>>> > > and discussion thread:
>>> > > https://www.mail-archive.com/dev@kafka.apache.org/msg86753.html
>>> > >
>>> > >
>>> > > Thanks,
>>> > >
>>> > > Anna
>>> > >
>>>
>>
>>
>
Re: [VOTE] KIP-279: Fix log divergence between leader and follower
after fast leader fail over
Posted by Anna Povzner <an...@confluent.io>.
Guozhang, thanks for catching this, I fixed the description (the example
assumed response with 21, '11' was a typo).
On Tue, Apr 17, 2018 at 1:25 PM, Anna Povzner <an...@confluent.io> wrote:
> Hi Colin,
>
> Yes, the impact of "losing" entries in the LeaderEpoch file is more
> round-trips for OffsetForLeaderEpoch. There is JIRA for the log cleaner:
> https://issues.apache.org/jira/browse/KAFKA-6780, and to investigate if
> there is actual possibility of losing committed records due to cleaning
> further than high watermark. In any case, this KIP does not make it any
> more or less likely.
>
> Thanks,
> Anna
>
>
>
> On Tue, Apr 17, 2018 at 11:53 AM, Colin McCabe <cm...@apache.org> wrote:
>
>> Thanks, Anna, this looks great.
>>
>> From the KIP:
>>
>> > Impact of topic compaction
>> >
>> > The proposed solution requires that we preserve history in
>> > LeaderEpochSequence file. Note that this is also required in the
>> current
>> > implementation if we want to guarantee no log divergence. The only
>> reason
>> > for "losing" entries in LeaderEpoch file is if we actually lose
>> > LeaderEpoch file and have to rebuild it from the log. If we delete all
>> > offsets for a particular epoch for some topic partition, we may miss
>> some
>> > entries in the LeaderEpochSequence file.
>> >
>> > We will not do any changes to compaction logic in this KIP, but here
>> is
>> > possible fixes to compaction logic:
>> >
>> > Leave a tombstone in the log if we delete all offsets for some
>> epoch,
>> > so that LeaderEpoch file can be rebuilt
>> > Do not compact further than persistent HW.
>>
>> Sorry if this has been answered before (I didn't find it in the DISCUSS
>> thread) but what is the impact of "losing" these entries in the LeaderEpoch
>> file? I suppose it would mean that more round-trips for
>> OffsetForLeaderEpoch might be required during a leader change. That's the
>> only impact, right?
>>
>> best,
>> Colin
>>
>> On Mon, Apr 16, 2018, at 09:48, Ismael Juma wrote:
>> > Thanks for the detailed KIP. +1 (binding)
>> >
>> > Ismael
>> >
>> > On Sat, Apr 14, 2018 at 3:54 PM, Anna Povzner <an...@confluent.io>
>> wrote:
>> >
>> > > Hi All,
>> > >
>> > >
>> > > I would like to start the vote on KIP-279: Fix log divergence between
>> > > leader and follower after fast leader fail over.
>> > >
>> > >
>> > > For reference, here's the KIP wiki:
>> > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
>> > > 279%3A+Fix+log+divergence+between+leader+and+follower+
>> > > after+fast+leader+fail+over
>> > >
>> > >
>> > >
>> > > and discussion thread:
>> > > https://www.mail-archive.com/dev@kafka.apache.org/msg86753.html
>> > >
>> > >
>> > > Thanks,
>> > >
>> > > Anna
>> > >
>>
>
>
Re: [VOTE] KIP-279: Fix log divergence between leader and follower
after fast leader fail over
Posted by Anna Povzner <an...@confluent.io>.
Hi Colin,
Yes, the impact of "losing" entries in the LeaderEpoch file is more
round-trips for OffsetForLeaderEpoch. There is JIRA for the log cleaner:
https://issues.apache.org/jira/browse/KAFKA-6780, and to investigate if
there is actual possibility of losing committed records due to cleaning
further than high watermark. In any case, this KIP does not make it any
more or less likely.
Thanks,
Anna
On Tue, Apr 17, 2018 at 11:53 AM, Colin McCabe <cm...@apache.org> wrote:
> Thanks, Anna, this looks great.
>
> From the KIP:
>
> > Impact of topic compaction
> >
> > The proposed solution requires that we preserve history in
> > LeaderEpochSequence file. Note that this is also required in the
> current
> > implementation if we want to guarantee no log divergence. The only
> reason
> > for "losing" entries in LeaderEpoch file is if we actually lose
> > LeaderEpoch file and have to rebuild it from the log. If we delete all
> > offsets for a particular epoch for some topic partition, we may miss
> some
> > entries in the LeaderEpochSequence file.
> >
> > We will not do any changes to compaction logic in this KIP, but here is
> > possible fixes to compaction logic:
> >
> > Leave a tombstone in the log if we delete all offsets for some
> epoch,
> > so that LeaderEpoch file can be rebuilt
> > Do not compact further than persistent HW.
>
> Sorry if this has been answered before (I didn't find it in the DISCUSS
> thread) but what is the impact of "losing" these entries in the LeaderEpoch
> file? I suppose it would mean that more round-trips for
> OffsetForLeaderEpoch might be required during a leader change. That's the
> only impact, right?
>
> best,
> Colin
>
> On Mon, Apr 16, 2018, at 09:48, Ismael Juma wrote:
> > Thanks for the detailed KIP. +1 (binding)
> >
> > Ismael
> >
> > On Sat, Apr 14, 2018 at 3:54 PM, Anna Povzner <an...@confluent.io> wrote:
> >
> > > Hi All,
> > >
> > >
> > > I would like to start the vote on KIP-279: Fix log divergence between
> > > leader and follower after fast leader fail over.
> > >
> > >
> > > For reference, here's the KIP wiki:
> > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> > > 279%3A+Fix+log+divergence+between+leader+and+follower+
> > > after+fast+leader+fail+over
> > >
> > >
> > >
> > > and discussion thread:
> > > https://www.mail-archive.com/dev@kafka.apache.org/msg86753.html
> > >
> > >
> > > Thanks,
> > >
> > > Anna
> > >
>
Re: [VOTE] KIP-279: Fix log divergence between leader and follower after
fast leader fail over
Posted by Colin McCabe <cm...@apache.org>.
Thanks, Anna, this looks great.
From the KIP:
> Impact of topic compaction
>
> The proposed solution requires that we preserve history in
> LeaderEpochSequence file. Note that this is also required in the current
> implementation if we want to guarantee no log divergence. The only reason
> for "losing" entries in LeaderEpoch file is if we actually lose
> LeaderEpoch file and have to rebuild it from the log. If we delete all
> offsets for a particular epoch for some topic partition, we may miss some
> entries in the LeaderEpochSequence file.
>
> We will not do any changes to compaction logic in this KIP, but here is
> possible fixes to compaction logic:
>
> Leave a tombstone in the log if we delete all offsets for some epoch,
> so that LeaderEpoch file can be rebuilt
> Do not compact further than persistent HW.
Sorry if this has been answered before (I didn't find it in the DISCUSS thread) but what is the impact of "losing" these entries in the LeaderEpoch file? I suppose it would mean that more round-trips for OffsetForLeaderEpoch might be required during a leader change. That's the only impact, right?
best,
Colin
On Mon, Apr 16, 2018, at 09:48, Ismael Juma wrote:
> Thanks for the detailed KIP. +1 (binding)
>
> Ismael
>
> On Sat, Apr 14, 2018 at 3:54 PM, Anna Povzner <an...@confluent.io> wrote:
>
> > Hi All,
> >
> >
> > I would like to start the vote on KIP-279: Fix log divergence between
> > leader and follower after fast leader fail over.
> >
> >
> > For reference, here's the KIP wiki:
> > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> > 279%3A+Fix+log+divergence+between+leader+and+follower+
> > after+fast+leader+fail+over
> >
> >
> >
> > and discussion thread:
> > https://www.mail-archive.com/dev@kafka.apache.org/msg86753.html
> >
> >
> > Thanks,
> >
> > Anna
> >
Re: [VOTE] KIP-279: Fix log divergence between leader and follower
after fast leader fail over
Posted by Ismael Juma <is...@juma.me.uk>.
Thanks for the detailed KIP. +1 (binding)
Ismael
On Sat, Apr 14, 2018 at 3:54 PM, Anna Povzner <an...@confluent.io> wrote:
> Hi All,
>
>
> I would like to start the vote on KIP-279: Fix log divergence between
> leader and follower after fast leader fail over.
>
>
> For reference, here's the KIP wiki:
> https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> 279%3A+Fix+log+divergence+between+leader+and+follower+
> after+fast+leader+fail+over
>
>
>
> and discussion thread:
> https://www.mail-archive.com/dev@kafka.apache.org/msg86753.html
>
>
> Thanks,
>
> Anna
>
Re: [VOTE] KIP-279: Fix log divergence between leader and follower
after fast leader fail over
Posted by Ben Stopford <be...@confluent.io>.
+1 thanks Anna
On Mon, Apr 16, 2018 at 5:37 PM Jason Gustafson <ja...@confluent.io> wrote:
> +1 Thanks for the KIP!
>
> On Sun, Apr 15, 2018 at 4:04 PM, Damian Guy <da...@gmail.com> wrote:
>
> > Thanks Anna +1
> > On Sun, 15 Apr 2018 at 15:40, Guozhang Wang <wa...@gmail.com> wrote:
> >
> > > Anna, thanks for the KIP! +1 from me.
> > >
> > > Just one minor comment: `Broker A will respond with offset 11`, seems
> it
> > > should be `21`.
> > >
> > > On Sun, Apr 15, 2018 at 3:48 AM, Dong Lin <li...@gmail.com> wrote:
> > >
> > > > Thanks for the KIP! LGTM. +1
> > > >
> > > > On Sat, Apr 14, 2018 at 5:31 PM, Ted Yu <yu...@gmail.com> wrote:
> > > >
> > > > > +1
> > > > >
> > > > > On Sat, Apr 14, 2018 at 3:54 PM, Anna Povzner <an...@confluent.io>
> > > wrote:
> > > > >
> > > > > > Hi All,
> > > > > >
> > > > > >
> > > > > > I would like to start the vote on KIP-279: Fix log divergence
> > between
> > > > > > leader and follower after fast leader fail over.
> > > > > >
> > > > > >
> > > > > > For reference, here's the KIP wiki:
> > > > > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> > > > > > 279%3A+Fix+log+divergence+between+leader+and+follower+
> > > > > > after+fast+leader+fail+over
> > > > > >
> > > > > >
> > > > > >
> > > > > > and discussion thread:
> > > > > > https://www.mail-archive.com/dev@kafka.apache.org/msg86753.html
> > > > > >
> > > > > >
> > > > > > Thanks,
> > > > > >
> > > > > > Anna
> > > > > >
> > > > >
> > > >
> > >
> > >
> > >
> > > --
> > > -- Guozhang
> > >
> >
>
Re: [VOTE] KIP-279: Fix log divergence between leader and follower
after fast leader fail over
Posted by Jason Gustafson <ja...@confluent.io>.
+1 Thanks for the KIP!
On Sun, Apr 15, 2018 at 4:04 PM, Damian Guy <da...@gmail.com> wrote:
> Thanks Anna +1
> On Sun, 15 Apr 2018 at 15:40, Guozhang Wang <wa...@gmail.com> wrote:
>
> > Anna, thanks for the KIP! +1 from me.
> >
> > Just one minor comment: `Broker A will respond with offset 11`, seems it
> > should be `21`.
> >
> > On Sun, Apr 15, 2018 at 3:48 AM, Dong Lin <li...@gmail.com> wrote:
> >
> > > Thanks for the KIP! LGTM. +1
> > >
> > > On Sat, Apr 14, 2018 at 5:31 PM, Ted Yu <yu...@gmail.com> wrote:
> > >
> > > > +1
> > > >
> > > > On Sat, Apr 14, 2018 at 3:54 PM, Anna Povzner <an...@confluent.io>
> > wrote:
> > > >
> > > > > Hi All,
> > > > >
> > > > >
> > > > > I would like to start the vote on KIP-279: Fix log divergence
> between
> > > > > leader and follower after fast leader fail over.
> > > > >
> > > > >
> > > > > For reference, here's the KIP wiki:
> > > > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> > > > > 279%3A+Fix+log+divergence+between+leader+and+follower+
> > > > > after+fast+leader+fail+over
> > > > >
> > > > >
> > > > >
> > > > > and discussion thread:
> > > > > https://www.mail-archive.com/dev@kafka.apache.org/msg86753.html
> > > > >
> > > > >
> > > > > Thanks,
> > > > >
> > > > > Anna
> > > > >
> > > >
> > >
> >
> >
> >
> > --
> > -- Guozhang
> >
>
Re: [VOTE] KIP-279: Fix log divergence between leader and follower
after fast leader fail over
Posted by Damian Guy <da...@gmail.com>.
Thanks Anna +1
On Sun, 15 Apr 2018 at 15:40, Guozhang Wang <wa...@gmail.com> wrote:
> Anna, thanks for the KIP! +1 from me.
>
> Just one minor comment: `Broker A will respond with offset 11`, seems it
> should be `21`.
>
> On Sun, Apr 15, 2018 at 3:48 AM, Dong Lin <li...@gmail.com> wrote:
>
> > Thanks for the KIP! LGTM. +1
> >
> > On Sat, Apr 14, 2018 at 5:31 PM, Ted Yu <yu...@gmail.com> wrote:
> >
> > > +1
> > >
> > > On Sat, Apr 14, 2018 at 3:54 PM, Anna Povzner <an...@confluent.io>
> wrote:
> > >
> > > > Hi All,
> > > >
> > > >
> > > > I would like to start the vote on KIP-279: Fix log divergence between
> > > > leader and follower after fast leader fail over.
> > > >
> > > >
> > > > For reference, here's the KIP wiki:
> > > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> > > > 279%3A+Fix+log+divergence+between+leader+and+follower+
> > > > after+fast+leader+fail+over
> > > >
> > > >
> > > >
> > > > and discussion thread:
> > > > https://www.mail-archive.com/dev@kafka.apache.org/msg86753.html
> > > >
> > > >
> > > > Thanks,
> > > >
> > > > Anna
> > > >
> > >
> >
>
>
>
> --
> -- Guozhang
>
Re: [VOTE] KIP-279: Fix log divergence between leader and follower
after fast leader fail over
Posted by Guozhang Wang <wa...@gmail.com>.
Anna, thanks for the KIP! +1 from me.
Just one minor comment: `Broker A will respond with offset 11`, seems it
should be `21`.
On Sun, Apr 15, 2018 at 3:48 AM, Dong Lin <li...@gmail.com> wrote:
> Thanks for the KIP! LGTM. +1
>
> On Sat, Apr 14, 2018 at 5:31 PM, Ted Yu <yu...@gmail.com> wrote:
>
> > +1
> >
> > On Sat, Apr 14, 2018 at 3:54 PM, Anna Povzner <an...@confluent.io> wrote:
> >
> > > Hi All,
> > >
> > >
> > > I would like to start the vote on KIP-279: Fix log divergence between
> > > leader and follower after fast leader fail over.
> > >
> > >
> > > For reference, here's the KIP wiki:
> > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> > > 279%3A+Fix+log+divergence+between+leader+and+follower+
> > > after+fast+leader+fail+over
> > >
> > >
> > >
> > > and discussion thread:
> > > https://www.mail-archive.com/dev@kafka.apache.org/msg86753.html
> > >
> > >
> > > Thanks,
> > >
> > > Anna
> > >
> >
>
--
-- Guozhang
Re: [VOTE] KIP-279: Fix log divergence between leader and follower
after fast leader fail over
Posted by Dong Lin <li...@gmail.com>.
Thanks for the KIP! LGTM. +1
On Sat, Apr 14, 2018 at 5:31 PM, Ted Yu <yu...@gmail.com> wrote:
> +1
>
> On Sat, Apr 14, 2018 at 3:54 PM, Anna Povzner <an...@confluent.io> wrote:
>
> > Hi All,
> >
> >
> > I would like to start the vote on KIP-279: Fix log divergence between
> > leader and follower after fast leader fail over.
> >
> >
> > For reference, here's the KIP wiki:
> > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> > 279%3A+Fix+log+divergence+between+leader+and+follower+
> > after+fast+leader+fail+over
> >
> >
> >
> > and discussion thread:
> > https://www.mail-archive.com/dev@kafka.apache.org/msg86753.html
> >
> >
> > Thanks,
> >
> > Anna
> >
>
Re: [VOTE] KIP-279: Fix log divergence between leader and follower
after fast leader fail over
Posted by Ted Yu <yu...@gmail.com>.
+1
On Sat, Apr 14, 2018 at 3:54 PM, Anna Povzner <an...@confluent.io> wrote:
> Hi All,
>
>
> I would like to start the vote on KIP-279: Fix log divergence between
> leader and follower after fast leader fail over.
>
>
> For reference, here's the KIP wiki:
> https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> 279%3A+Fix+log+divergence+between+leader+and+follower+
> after+fast+leader+fail+over
>
>
>
> and discussion thread:
> https://www.mail-archive.com/dev@kafka.apache.org/msg86753.html
>
>
> Thanks,
>
> Anna
>