Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

Dong,

Thanks for the comments.   I have updated the KIP based on your comments.

below is reply to your questions:

1.  We only calculate this metric for log compaction that is determined by
max compaction lag. So we only collect non-negative metrics.  The log
cleaner is consistently running with some back off time if no job needs to
be done.
The max is the max among all log cleaner threads in their latest run not
the historical max.  This is similar to existing metric
"max-clean-time-secs".  I now mentioned this is metric from each thread in
the KIP.
User can look at the historical data to track how delay changes over time
(similar as other log cleaner metrics).

Another way of defining this metric is : "compaction_finish_time -
earliest_timestamp_of_first_uncompacted_segment".  So it is not w.r.t.
However,  max compaction lag may vary for different topics, and this
doesn't really tell how soon a compaction request is fulfilled after max
compaction lag.  What do you think?

2.  This is intent to track whether the latest logs compacted are
determined by max compaction lag.
The metric will be updated for each log cleaner run. If there are 2 two log
cleaner threads, and they both work on log partitions determined by "max
compaction lag" in their last run,  the value of this metric will be 2.
The previous metric doesn't provide this information if there are more than
one log cleaner thread.

3. I meant to say it is required to be picked up by log compaction after
this max lag. But the actual compaction finish time may vary, since the log
cleaner may take time to finish compaction on this partition or log cleaner
may work on other partition first.
Guarantee may be misleading, I have updated the KIP.

4. It is determined based on the cleaner checkpoint file.  This KIP doesn't
change how broker determined the un-compacted segments.
5.  done.
6.  Why should we need to make this feature depends upon message
timestamp?  "segment.largestTimestamp - maxSegmentMs" is
a reasonable estimate to determine the violation of max compaction lag,
and this estimate is only needed if the first segment of a log partition is
un-compacted.
7.  I removed unrelated part, and specifically mentioned the added
metric "num-logs-compacted-by-max-compaction-lag"
can be used for this performance impact measurement.

Xiongqi (Wesley) Wu


On Tue, Nov 6, 2018 at 6:50 PM Dong Lin  wrote:

> Hey Xiongqi,
>
> Thanks for the update. A few more comments below
>
> 1) According to the definition of
> kafka.log:type=LogCleaner,name=max-compaction-delay, it seems that the
> metric value will be a large negative number if max.compaction.lag.ms is
> MAX_LONG. Would this be a problem? Also, it seems weird that the value of
> the metric is defined w.r.t. how often the log cleaner is run.
>
> 2) Not sure if we need the metric num-logs-compacted-by-max-compaction-lag
> in addition to max-compaction-delay. It seems that operator can just use
> max-compaction-delay to determine whether the max.compaction.lag is
> properly enforced in a quantitative manner. Also, the metric name
> `num-logs-compacted-by-max-compaction-lag` is inconsistent with its
> intended meaning, i.e. the number of logs that needs to be compacted due to
> max.compaction.lag but not yet compacted. So it is probably simple to just
> remove this metric.
>
> 3) The KIP currently says that "a message record has a guaranteed
> upper-bound in time to become mandatory for compaction". The word
> "guarantee" may be misleading because the message may still not be
> compacted within max.compaction.lag after its creation. Could you clarify
> the exact semantics of the max.compaction.lag.ms in the Public Interface
> section?
>
> 4) The KIP's proposed change will estimate earliest message timestamp for
> un-compacted log segments. Can you explain how broker determines whether a
> segment has been compacted after the broker is restarted?
>
> 5) 2.b in Proposed Change section provides two way to get timestamp. To
> make the KIP easier to read for future reference, could we just mention the
> method that we plan to use and move the other solution to the rejected
> alternative section?
>
> 6) Based on the discussion (i.e. point 2 in the previous email), it is said
> that we can assume all messages have timestamp and the feature added in
> this KIP can be skipped for those messages which do not have timestamp. So
> do we still need to use "segment.largestTimestamp - maxSegmentMs" in
> Proposed Change section 2.a?
>
> 7) Based on the discussion (i.e. point 8 in the previous email), if this
> KIP requires user to monitor certain existing metrics for performance
> impact added in this KIP, can we list the metrics in the KIP for user's
> convenience?
>
>
> Thanks,
> Dong
>
> On Mon, Oct 29, 2018 at 3:16 PM xiongqi wu  wrote:
>
> > Hi Dong,
> > I have updated the KIP to address your comments.
> > One correction to previous Email:
> > after offline discussion with Dong,  we decide to use MAX_LONG as default
> > 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Dong Lin]

Hey Xiongqi,

Thanks for the update. A few more comments below

1) According to the definition of
kafka.log:type=LogCleaner,name=max-compaction-delay, it seems that the
metric value will be a large negative number if max.compaction.lag.ms is
MAX_LONG. Would this be a problem? Also, it seems weird that the value of
the metric is defined w.r.t. how often the log cleaner is run.

2) Not sure if we need the metric num-logs-compacted-by-max-compaction-lag
in addition to max-compaction-delay. It seems that operator can just use
max-compaction-delay to determine whether the max.compaction.lag is
properly enforced in a quantitative manner. Also, the metric name
`num-logs-compacted-by-max-compaction-lag` is inconsistent with its
intended meaning, i.e. the number of logs that needs to be compacted due to
max.compaction.lag but not yet compacted. So it is probably simple to just
remove this metric.

3) The KIP currently says that "a message record has a guaranteed
upper-bound in time to become mandatory for compaction". The word
"guarantee" may be misleading because the message may still not be
compacted within max.compaction.lag after its creation. Could you clarify
the exact semantics of the max.compaction.lag.ms in the Public Interface
section?

4) The KIP's proposed change will estimate earliest message timestamp for
un-compacted log segments. Can you explain how broker determines whether a
segment has been compacted after the broker is restarted?

5) 2.b in Proposed Change section provides two way to get timestamp. To
make the KIP easier to read for future reference, could we just mention the
method that we plan to use and move the other solution to the rejected
alternative section?

6) Based on the discussion (i.e. point 2 in the previous email), it is said
that we can assume all messages have timestamp and the feature added in
this KIP can be skipped for those messages which do not have timestamp. So
do we still need to use "segment.largestTimestamp - maxSegmentMs" in
Proposed Change section 2.a?

7) Based on the discussion (i.e. point 8 in the previous email), if this
KIP requires user to monitor certain existing metrics for performance
impact added in this KIP, can we list the metrics in the KIP for user's
convenience?


Thanks,
Dong

On Mon, Oct 29, 2018 at 3:16 PM xiongqi wu  wrote:

> Hi Dong,
> I have updated the KIP to address your comments.
> One correction to previous Email:
> after offline discussion with Dong,  we decide to use MAX_LONG as default
> value for max.compaction.lag.ms.
>
>
> Xiongqi (Wesley) Wu
>
>
> On Mon, Oct 29, 2018 at 12:15 PM xiongqi wu  wrote:
>
> > Hi Dong,
> >
> > Thank you for your comment.  See my inline comments.
> > I will update the KIP shortly.
> >
> > Xiongqi (Wesley) Wu
> >
> >
> > On Sun, Oct 28, 2018 at 9:17 PM Dong Lin  wrote:
> >
> >> Hey Xiongqi,
> >>
> >> Sorry for late reply. I have some comments below:
> >>
> >> 1) As discussed earlier in the email list, if the topic is configured
> with
> >> both deletion and compaction, in some cases messages produced a long
> time
> >> ago can not be deleted based on time. This is a valid use-case because
> we
> >> actually have topic which is configured with both deletion and
> compaction
> >> policy. And we should enforce the semantics for both policy. Solution A
> >> sounds good. We do not need interface change (e.g. extra config) to
> >> enforce
> >> solution A. All we need is to update implementation so that when broker
> >> compacts a topic, if the message has timestamp (which is the common
> case),
> >> messages that are too old (based on the time-based retention config)
> will
> >> be discarded. Since this is a valid issue and it is also related to the
> >> guarantee of when a message can be deleted, can we include the solution
> of
> >> this problem in the KIP?
> >>
> > ==  This makes sense.  We can use similar approach to increase the
> log
> > start offset.
> >
> >>
> >> 2) It is probably OK to assume that all messages have timestamp. The
> >> per-message timestamp was introduced into Kafka 0.10.0 with KIP-31 and
> >> KIP-32 as of Feb 2016. Kafka 0.10.0 or earlier versions are no longer
> >> supported. Also, since the use-case for this feature is primarily for
> >> GDPR,
> >> we can assume that client library has already been upgraded to support
> >> SSL,
> >> which feature is added after KIP-31 and KIP-32.
> >>
> >>  =>  Ok. We can use message timestamp to delete expired records
> > if both compaction and retention are enabled.
> >
> >
> > 3) In Proposed Change section 2.a, it is said that
> segment.largestTimestamp
> >> - maxSegmentMs can be used to determine the timestamp of the earliest
> >> message. Would it be simpler to just use the create time of the file to
> >> determine the time?
> >>
> >> >  Linux/Java doesn't provide API for file creation time because
> > some filesystem type doesn't provide file creation time.
> >
> >
> >> 4) The KIP suggests to use must-clean-ratio to select the 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

Hi Dong,
I have updated the KIP to address your comments.
One correction to previous Email:
after offline discussion with Dong,  we decide to use MAX_LONG as default
value for max.compaction.lag.ms.


Xiongqi (Wesley) Wu


On Mon, Oct 29, 2018 at 12:15 PM xiongqi wu  wrote:

> Hi Dong,
>
> Thank you for your comment.  See my inline comments.
> I will update the KIP shortly.
>
> Xiongqi (Wesley) Wu
>
>
> On Sun, Oct 28, 2018 at 9:17 PM Dong Lin  wrote:
>
>> Hey Xiongqi,
>>
>> Sorry for late reply. I have some comments below:
>>
>> 1) As discussed earlier in the email list, if the topic is configured with
>> both deletion and compaction, in some cases messages produced a long time
>> ago can not be deleted based on time. This is a valid use-case because we
>> actually have topic which is configured with both deletion and compaction
>> policy. And we should enforce the semantics for both policy. Solution A
>> sounds good. We do not need interface change (e.g. extra config) to
>> enforce
>> solution A. All we need is to update implementation so that when broker
>> compacts a topic, if the message has timestamp (which is the common case),
>> messages that are too old (based on the time-based retention config) will
>> be discarded. Since this is a valid issue and it is also related to the
>> guarantee of when a message can be deleted, can we include the solution of
>> this problem in the KIP?
>>
> ==  This makes sense.  We can use similar approach to increase the log
> start offset.
>
>>
>> 2) It is probably OK to assume that all messages have timestamp. The
>> per-message timestamp was introduced into Kafka 0.10.0 with KIP-31 and
>> KIP-32 as of Feb 2016. Kafka 0.10.0 or earlier versions are no longer
>> supported. Also, since the use-case for this feature is primarily for
>> GDPR,
>> we can assume that client library has already been upgraded to support
>> SSL,
>> which feature is added after KIP-31 and KIP-32.
>>
>>  =>  Ok. We can use message timestamp to delete expired records
> if both compaction and retention are enabled.
>
>
> 3) In Proposed Change section 2.a, it is said that segment.largestTimestamp
>> - maxSegmentMs can be used to determine the timestamp of the earliest
>> message. Would it be simpler to just use the create time of the file to
>> determine the time?
>>
>> >  Linux/Java doesn't provide API for file creation time because
> some filesystem type doesn't provide file creation time.
>
>
>> 4) The KIP suggests to use must-clean-ratio to select the partition to be
>> compacted. Unlike dirty ratio which is mostly for performance, the logs
>> whose "must-clean-ratio" is non-zero must be compacted immediately for
>> correctness reason (and for GDPR). And if this can no be achieved because
>> e.g. broker compaction throughput is too low, investigation will be
>> needed.
>> So it seems simpler to first compact logs which has segment whose earliest
>> timetamp is earlier than now - max.compaction.lag.ms, instead of defining
>> must-clean-ratio and sorting logs based on this value.
>>
>>
> ==>  Good suggestion. This can simply the implementation quite a bit
> if we are not too concerned about compaction of GDPR required partition
> queued behind some large partition.  The actual compaction completion time
> is not guaranteed anyway.
>
>
>> 5) The KIP says max.compaction.lag.ms is 0 by default and it is also
>> suggested that 0 means disable. Should we set this value to MAX_LONG by
>> default to effectively disable the feature added in this KIP?
>>
>> > I would rather use 0 so the corresponding code path will not be
> exercised.  By using MAX_LONG, we would theoretically go through related
> code to find out whether the partition is required to be compacted to
> satisfy MAX_LONG.
>
> 6) It is probably cleaner and readable not to include in Public Interface
>> section those configs whose meaning is not changed.
>>
>> > I will clean that up.
>
> 7) The goal of this KIP is to ensure that log segment whose earliest
>> message is earlier than a given threshold will be compacted. This goal may
>> not be achieved if the compact throughput can not catchup with the total
>> bytes-in-rate for the compacted topics on the broker. Thus we need an easy
>> way to tell operator whether this goal is achieved. If we don't already
>> have such metric, maybe we can include metrics to show 1) the total number
>> of log segments (or logs) which needs to be immediately compacted as
>> determined by max.compaction.lag; and 2) the maximum value of now -
>> earliest_time_stamp_of_segment among all segments that needs to be
>> compacted.
>>
>> ===> good suggestion.  I will update KIP for these metrics.
>
> 8) The Performance Impact suggests user to use the existing metrics to
>> monitor the performance impact of this KIP. It i useful to list mean of
>> each jmx metrics that we want user to monitor, and possibly explain how to
>> interpret the value of these metrics to determine whether 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

Hi Dong,

Thank you for your comment.  See my inline comments.
I will update the KIP shortly.

Xiongqi (Wesley) Wu


On Sun, Oct 28, 2018 at 9:17 PM Dong Lin  wrote:

> Hey Xiongqi,
>
> Sorry for late reply. I have some comments below:
>
> 1) As discussed earlier in the email list, if the topic is configured with
> both deletion and compaction, in some cases messages produced a long time
> ago can not be deleted based on time. This is a valid use-case because we
> actually have topic which is configured with both deletion and compaction
> policy. And we should enforce the semantics for both policy. Solution A
> sounds good. We do not need interface change (e.g. extra config) to enforce
> solution A. All we need is to update implementation so that when broker
> compacts a topic, if the message has timestamp (which is the common case),
> messages that are too old (based on the time-based retention config) will
> be discarded. Since this is a valid issue and it is also related to the
> guarantee of when a message can be deleted, can we include the solution of
> this problem in the KIP?
>
==  This makes sense.  We can use similar approach to increase the log
start offset.

>
> 2) It is probably OK to assume that all messages have timestamp. The
> per-message timestamp was introduced into Kafka 0.10.0 with KIP-31 and
> KIP-32 as of Feb 2016. Kafka 0.10.0 or earlier versions are no longer
> supported. Also, since the use-case for this feature is primarily for GDPR,
> we can assume that client library has already been upgraded to support SSL,
> which feature is added after KIP-31 and KIP-32.
>
>  =>  Ok. We can use message timestamp to delete expired records if
both compaction and retention are enabled.


3) In Proposed Change section 2.a, it is said that segment.largestTimestamp
> - maxSegmentMs can be used to determine the timestamp of the earliest
> message. Would it be simpler to just use the create time of the file to
> determine the time?
>
> >  Linux/Java doesn't provide API for file creation time because
some filesystem type doesn't provide file creation time.


> 4) The KIP suggests to use must-clean-ratio to select the partition to be
> compacted. Unlike dirty ratio which is mostly for performance, the logs
> whose "must-clean-ratio" is non-zero must be compacted immediately for
> correctness reason (and for GDPR). And if this can no be achieved because
> e.g. broker compaction throughput is too low, investigation will be needed.
> So it seems simpler to first compact logs which has segment whose earliest
> timetamp is earlier than now - max.compaction.lag.ms, instead of defining
> must-clean-ratio and sorting logs based on this value.
>
>
==>  Good suggestion. This can simply the implementation quite a bit if
we are not too concerned about compaction of GDPR required partition queued
behind some large partition.  The actual compaction completion time is not
guaranteed anyway.


> 5) The KIP says max.compaction.lag.ms is 0 by default and it is also
> suggested that 0 means disable. Should we set this value to MAX_LONG by
> default to effectively disable the feature added in this KIP?
>
> > I would rather use 0 so the corresponding code path will not be
exercised.  By using MAX_LONG, we would theoretically go through related
code to find out whether the partition is required to be compacted to
satisfy MAX_LONG.

6) It is probably cleaner and readable not to include in Public Interface
> section those configs whose meaning is not changed.
>
> > I will clean that up.

7) The goal of this KIP is to ensure that log segment whose earliest
> message is earlier than a given threshold will be compacted. This goal may
> not be achieved if the compact throughput can not catchup with the total
> bytes-in-rate for the compacted topics on the broker. Thus we need an easy
> way to tell operator whether this goal is achieved. If we don't already
> have such metric, maybe we can include metrics to show 1) the total number
> of log segments (or logs) which needs to be immediately compacted as
> determined by max.compaction.lag; and 2) the maximum value of now -
> earliest_time_stamp_of_segment among all segments that needs to be
> compacted.
>
> ===> good suggestion.  I will update KIP for these metrics.

8) The Performance Impact suggests user to use the existing metrics to
> monitor the performance impact of this KIP. It i useful to list mean of
> each jmx metrics that we want user to monitor, and possibly explain how to
> interpret the value of these metrics to determine whether there is
> performance issue.
>
> =>  I will update the KIP.

> Thanks,
> Dong
>
> On Tue, Oct 16, 2018 at 10:53 AM xiongqi wu  wrote:
>
> > Mayuresh,
> >
> > Thanks for the comments.
> > The requirement is that we need to pick up segments that are older than
> > maxCompactionLagMs for compaction.
> > maxCompactionLagMs is an upper-bound, which implies that picking up
> > segments for compaction 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Dong Lin]

Hey Xiongqi,

Sorry for late reply. I have some comments below:

1) As discussed earlier in the email list, if the topic is configured with
both deletion and compaction, in some cases messages produced a long time
ago can not be deleted based on time. This is a valid use-case because we
actually have topic which is configured with both deletion and compaction
policy. And we should enforce the semantics for both policy. Solution A
sounds good. We do not need interface change (e.g. extra config) to enforce
solution A. All we need is to update implementation so that when broker
compacts a topic, if the message has timestamp (which is the common case),
messages that are too old (based on the time-based retention config) will
be discarded. Since this is a valid issue and it is also related to the
guarantee of when a message can be deleted, can we include the solution of
this problem in the KIP?

2) It is probably OK to assume that all messages have timestamp. The
per-message timestamp was introduced into Kafka 0.10.0 with KIP-31 and
KIP-32 as of Feb 2016. Kafka 0.10.0 or earlier versions are no longer
supported. Also, since the use-case for this feature is primarily for GDPR,
we can assume that client library has already been upgraded to support SSL,
which feature is added after KIP-31 and KIP-32.

3) In Proposed Change section 2.a, it is said that segment.largestTimestamp
- maxSegmentMs can be used to determine the timestamp of the earliest
message. Would it be simpler to just use the create time of the file to
determine the time?

4) The KIP suggests to use must-clean-ratio to select the partition to be
compacted. Unlike dirty ratio which is mostly for performance, the logs
whose "must-clean-ratio" is non-zero must be compacted immediately for
correctness reason (and for GDPR). And if this can no be achieved because
e.g. broker compaction throughput is too low, investigation will be needed.
So it seems simpler to first compact logs which has segment whose earliest
timetamp is earlier than now - max.compaction.lag.ms, instead of defining
must-clean-ratio and sorting logs based on this value.

5) The KIP says max.compaction.lag.ms is 0 by default and it is also
suggested that 0 means disable. Should we set this value to MAX_LONG by
default to effectively disable the feature added in this KIP?

6) It is probably cleaner and readable not to include in Public Interface
section those configs whose meaning is not changed.

7) The goal of this KIP is to ensure that log segment whose earliest
message is earlier than a given threshold will be compacted. This goal may
not be achieved if the compact throughput can not catchup with the total
bytes-in-rate for the compacted topics on the broker. Thus we need an easy
way to tell operator whether this goal is achieved. If we don't already
have such metric, maybe we can include metrics to show 1) the total number
of log segments (or logs) which needs to be immediately compacted as
determined by max.compaction.lag; and 2) the maximum value of now -
earliest_time_stamp_of_segment among all segments that needs to be
compacted.

8) The Performance Impact suggests user to use the existing metrics to
monitor the performance impact of this KIP. It i useful to list mean of
each jmx metrics that we want user to monitor, and possibly explain how to
interpret the value of these metrics to determine whether there is
performance issue.

Thanks,
Dong

On Tue, Oct 16, 2018 at 10:53 AM xiongqi wu  wrote:

> Mayuresh,
>
> Thanks for the comments.
> The requirement is that we need to pick up segments that are older than
> maxCompactionLagMs for compaction.
> maxCompactionLagMs is an upper-bound, which implies that picking up
> segments for compaction earlier doesn't violated the policy.
> We use the creation time of a segment as an estimation of its records
> arrival time, so these records can be compacted no later than
> maxCompactionLagMs.
>
> On the other hand, compaction is an expensive operation, we don't want to
> compact the log partition whenever a new segment is sealed.
> Therefore, we want to pick up a segment for compaction when the segment is
> closed to mandatory max compaction lag (so we use segment creation time as
> an estimation.)
>
>
> Xiongqi (Wesley) Wu
>
>
> On Mon, Oct 15, 2018 at 5:54 PM Mayuresh Gharat <
> gharatmayures...@gmail.com>
> wrote:
>
> > Hi Wesley,
> >
> > Thanks for the KIP and sorry for being late to the party.
> >  I wanted to understand, the scenario you mentioned in Proposed changes :
> >
> > -
> > >
> > > Estimate the earliest message timestamp of an un-compacted log segment.
> > we
> > > only need to estimate earliest message timestamp for un-compacted log
> > > segments to ensure timely compaction because the deletion requests that
> > > belong to compacted segments have already been processed.
> > >
> > >1.
> > >
> > >for the first (earliest) log segment:  The estimated earliest
> > >timestamp is set to the timestamp of the first message 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

Mayuresh,

Thanks for the comments.
The requirement is that we need to pick up segments that are older than
maxCompactionLagMs for compaction.
maxCompactionLagMs is an upper-bound, which implies that picking up
segments for compaction earlier doesn't violated the policy.
We use the creation time of a segment as an estimation of its records
arrival time, so these records can be compacted no later than
maxCompactionLagMs.

On the other hand, compaction is an expensive operation, we don't want to
compact the log partition whenever a new segment is sealed.
Therefore, we want to pick up a segment for compaction when the segment is
closed to mandatory max compaction lag (so we use segment creation time as
an estimation.)


Xiongqi (Wesley) Wu


On Mon, Oct 15, 2018 at 5:54 PM Mayuresh Gharat 
wrote:

> Hi Wesley,
>
> Thanks for the KIP and sorry for being late to the party.
>  I wanted to understand, the scenario you mentioned in Proposed changes :
>
> -
> >
> > Estimate the earliest message timestamp of an un-compacted log segment.
> we
> > only need to estimate earliest message timestamp for un-compacted log
> > segments to ensure timely compaction because the deletion requests that
> > belong to compacted segments have already been processed.
> >
> >1.
> >
> >for the first (earliest) log segment:  The estimated earliest
> >timestamp is set to the timestamp of the first message if timestamp is
> >present in the message. Otherwise, the estimated earliest timestamp
> is set
> >to "segment.largestTimestamp - maxSegmentMs”
> > (segment.largestTimestamp is lastModified time of the log segment or
> max
> >timestamp we see for the log segment.). In the later case, the actual
> >timestamp of the first message might be later than the estimation,
> but it
> >is safe to pick up the log for compaction earlier.
> >
> > When we say "actual timestamp of the first message might be later than
> the
> estimation, but it is safe to pick up the log for compaction earlier.",
> doesn't that violate the assumption that we will consider a segment for
> compaction only if the time of creation the segment has crossed the "now -
> maxCompactionLagMs" ?
>
> Thanks,
>
> Mayuresh
>
> On Mon, Sep 3, 2018 at 7:28 PM Brett Rann 
> wrote:
>
> > Might also be worth moving to a vote thread? Discussion seems to have
> gone
> > as far as it can.
> >
> > > On 4 Sep 2018, at 12:08, xiongqi wu  wrote:
> > >
> > > Brett,
> > >
> > > Yes, I will post PR tomorrow.
> > >
> > > Xiongqi (Wesley) Wu
> > >
> > >
> > > On Sun, Sep 2, 2018 at 6:28 PM Brett Rann 
> > wrote:
> > >
> > > > +1 (non-binding) from me on the interface. I'd like to see someone
> > familiar
> > > > with
> > > > the code comment on the approach, and note there's a couple of
> > different
> > > > approaches: what's documented in the KIP, and what Xiaohe Dong was
> > working
> > > > on
> > > > here:
> > > >
> > > >
> >
> https://github.com/dongxiaohe/kafka/tree/dongxiaohe/log-cleaner-compaction-max-lifetime-2.0
> > > >
> > > > If you have code working already Xiongqi Wu could you share a PR? I'd
> > be
> > > > happy
> > > > to start testing.
> > > >
> > > > On Tue, Aug 28, 2018 at 5:57 AM xiongqi wu 
> > wrote:
> > > >
> > > > > Hi All,
> > > > >
> > > > > Do you have any additional comments on this KIP?
> > > > >
> > > > >
> > > > > On Thu, Aug 16, 2018 at 9:17 PM, xiongqi wu 
> > wrote:
> > > > >
> > > > > > on 2)
> > > > > > The offsetmap is built starting from dirty segment.
> > > > > > The compaction starts from the beginning of the log partition.
> > That's
> > > > how
> > > > > > it ensure the deletion of tomb keys.
> > > > > > I will double check tomorrow.
> > > > > >
> > > > > > Xiongqi (Wesley) Wu
> > > > > >
> > > > > >
> > > > > > On Thu, Aug 16, 2018 at 6:46 PM Brett Rann
> > 
> > > > > > wrote:
> > > > > >
> > > > > >> To just clarify a bit on 1. whether there's an external
> storage/DB
> > > > isn't
> > > > > >> relevant here.
> > > > > >> Compacted topics allow a tombstone record to be sent (a null
> value
> > > > for a
> > > > > >> key) which
> > > > > >> currently will result in old values for that key being deleted
> if
> > some
> > > > > >> conditions are met.
> > > > > >> There are existing controls to make sure the old values will
> stay
> > > > around
> > > > > >> for a minimum
> > > > > >> time at least, but no dedicated control to ensure the tombstone
> > will
> > > > > >> delete
> > > > > >> within a
> > > > > >> maximum time.
> > > > > >>
> > > > > >> One popular reason that maximum time for deletion is desirable
> > right
> > > > now
> > > > > >> is
> > > > > >> GDPR with
> > > > > >> PII. But we're not proposing any GDPR awareness in kafka, just
> > being
> > > > > able
> > > > > >> to guarantee
> > > > > >> a max time where a tombstoned key will be removed from the
> > compacted
> > > > > >> topic.
> > > > > >>
> > > > > >> on 2)
> > > > > >> huh, i thought it kept track of the first dirty segment and
> didn't
> > > > > >> 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Mayuresh Gharat]

Hi Wesley,

Thanks for the KIP and sorry for being late to the party.
 I wanted to understand, the scenario you mentioned in Proposed changes :

-
>
> Estimate the earliest message timestamp of an un-compacted log segment. we
> only need to estimate earliest message timestamp for un-compacted log
> segments to ensure timely compaction because the deletion requests that
> belong to compacted segments have already been processed.
>
>1.
>
>for the first (earliest) log segment:  The estimated earliest
>timestamp is set to the timestamp of the first message if timestamp is
>present in the message. Otherwise, the estimated earliest timestamp is set
>to "segment.largestTimestamp - maxSegmentMs”
> (segment.largestTimestamp is lastModified time of the log segment or max
>timestamp we see for the log segment.). In the later case, the actual
>timestamp of the first message might be later than the estimation, but it
>is safe to pick up the log for compaction earlier.
>
> When we say "actual timestamp of the first message might be later than the
estimation, but it is safe to pick up the log for compaction earlier.",
doesn't that violate the assumption that we will consider a segment for
compaction only if the time of creation the segment has crossed the "now -
maxCompactionLagMs" ?

Thanks,

Mayuresh

On Mon, Sep 3, 2018 at 7:28 PM Brett Rann  wrote:

> Might also be worth moving to a vote thread? Discussion seems to have gone
> as far as it can.
>
> > On 4 Sep 2018, at 12:08, xiongqi wu  wrote:
> >
> > Brett,
> >
> > Yes, I will post PR tomorrow.
> >
> > Xiongqi (Wesley) Wu
> >
> >
> > On Sun, Sep 2, 2018 at 6:28 PM Brett Rann 
> wrote:
> >
> > > +1 (non-binding) from me on the interface. I'd like to see someone
> familiar
> > > with
> > > the code comment on the approach, and note there's a couple of
> different
> > > approaches: what's documented in the KIP, and what Xiaohe Dong was
> working
> > > on
> > > here:
> > >
> > >
> https://github.com/dongxiaohe/kafka/tree/dongxiaohe/log-cleaner-compaction-max-lifetime-2.0
> > >
> > > If you have code working already Xiongqi Wu could you share a PR? I'd
> be
> > > happy
> > > to start testing.
> > >
> > > On Tue, Aug 28, 2018 at 5:57 AM xiongqi wu 
> wrote:
> > >
> > > > Hi All,
> > > >
> > > > Do you have any additional comments on this KIP?
> > > >
> > > >
> > > > On Thu, Aug 16, 2018 at 9:17 PM, xiongqi wu 
> wrote:
> > > >
> > > > > on 2)
> > > > > The offsetmap is built starting from dirty segment.
> > > > > The compaction starts from the beginning of the log partition.
> That's
> > > how
> > > > > it ensure the deletion of tomb keys.
> > > > > I will double check tomorrow.
> > > > >
> > > > > Xiongqi (Wesley) Wu
> > > > >
> > > > >
> > > > > On Thu, Aug 16, 2018 at 6:46 PM Brett Rann
> 
> > > > > wrote:
> > > > >
> > > > >> To just clarify a bit on 1. whether there's an external storage/DB
> > > isn't
> > > > >> relevant here.
> > > > >> Compacted topics allow a tombstone record to be sent (a null value
> > > for a
> > > > >> key) which
> > > > >> currently will result in old values for that key being deleted if
> some
> > > > >> conditions are met.
> > > > >> There are existing controls to make sure the old values will stay
> > > around
> > > > >> for a minimum
> > > > >> time at least, but no dedicated control to ensure the tombstone
> will
> > > > >> delete
> > > > >> within a
> > > > >> maximum time.
> > > > >>
> > > > >> One popular reason that maximum time for deletion is desirable
> right
> > > now
> > > > >> is
> > > > >> GDPR with
> > > > >> PII. But we're not proposing any GDPR awareness in kafka, just
> being
> > > > able
> > > > >> to guarantee
> > > > >> a max time where a tombstoned key will be removed from the
> compacted
> > > > >> topic.
> > > > >>
> > > > >> on 2)
> > > > >> huh, i thought it kept track of the first dirty segment and didn't
> > > > >> recompact older "clean" ones.
> > > > >> But I didn't look at code or test for that.
> > > > >>
> > > > >> On Fri, Aug 17, 2018 at 10:57 AM xiongqi wu 
> > > > wrote:
> > > > >>
> > > > >> > 1, Owner of data (in this sense, kafka is the not the owner of
> data)
> > > > >> > should keep track of lifecycle of the data in some external
> > > > storage/DB.
> > > > >> > The owner determines when to delete the data and send the delete
> > > > >> request to
> > > > >> > kafka. Kafka doesn't know about the content of data but to
> provide a
> > > > >> mean
> > > > >> > for deletion.
> > > > >> >
> > > > >> > 2 , each time compaction runs, it will start from first
> segments (no
> > > > >> > matter if it is compacted or not). The time estimation here is
> only
> > > > used
> > > > >> > to determine whether we should run compaction on this log
> partition.
> > > > So
> > > > >> we
> > > > >> > only need to estimate uncompacted segments.
> > > > >> >
> > > > >> > On Thu, Aug 16, 2018 at 5:35 PM, Dong Lin 
> > > > wrote:
> > > > >> >
> > > > >> > > Hey Xiongqi,
> > > > >> > >

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Brett Rann]

Might also be worth moving to a vote thread? Discussion seems to have gone as 
far as it can. 

> On 4 Sep 2018, at 12:08, xiongqi wu  wrote:
> 
> Brett,
> 
> Yes, I will post PR tomorrow.
> 
> Xiongqi (Wesley) Wu
> 
> 
> On Sun, Sep 2, 2018 at 6:28 PM Brett Rann  wrote:
> 
> > +1 (non-binding) from me on the interface. I'd like to see someone familiar
> > with
> > the code comment on the approach, and note there's a couple of different
> > approaches: what's documented in the KIP, and what Xiaohe Dong was working
> > on
> > here:
> >
> > https://github.com/dongxiaohe/kafka/tree/dongxiaohe/log-cleaner-compaction-max-lifetime-2.0
> >
> > If you have code working already Xiongqi Wu could you share a PR? I'd be
> > happy
> > to start testing.
> >
> > On Tue, Aug 28, 2018 at 5:57 AM xiongqi wu  wrote:
> >
> > > Hi All,
> > >
> > > Do you have any additional comments on this KIP?
> > >
> > >
> > > On Thu, Aug 16, 2018 at 9:17 PM, xiongqi wu  wrote:
> > >
> > > > on 2)
> > > > The offsetmap is built starting from dirty segment.
> > > > The compaction starts from the beginning of the log partition. That's
> > how
> > > > it ensure the deletion of tomb keys.
> > > > I will double check tomorrow.
> > > >
> > > > Xiongqi (Wesley) Wu
> > > >
> > > >
> > > > On Thu, Aug 16, 2018 at 6:46 PM Brett Rann 
> > > > wrote:
> > > >
> > > >> To just clarify a bit on 1. whether there's an external storage/DB
> > isn't
> > > >> relevant here.
> > > >> Compacted topics allow a tombstone record to be sent (a null value
> > for a
> > > >> key) which
> > > >> currently will result in old values for that key being deleted if some
> > > >> conditions are met.
> > > >> There are existing controls to make sure the old values will stay
> > around
> > > >> for a minimum
> > > >> time at least, but no dedicated control to ensure the tombstone will
> > > >> delete
> > > >> within a
> > > >> maximum time.
> > > >>
> > > >> One popular reason that maximum time for deletion is desirable right
> > now
> > > >> is
> > > >> GDPR with
> > > >> PII. But we're not proposing any GDPR awareness in kafka, just being
> > > able
> > > >> to guarantee
> > > >> a max time where a tombstoned key will be removed from the compacted
> > > >> topic.
> > > >>
> > > >> on 2)
> > > >> huh, i thought it kept track of the first dirty segment and didn't
> > > >> recompact older "clean" ones.
> > > >> But I didn't look at code or test for that.
> > > >>
> > > >> On Fri, Aug 17, 2018 at 10:57 AM xiongqi wu 
> > > wrote:
> > > >>
> > > >> > 1, Owner of data (in this sense, kafka is the not the owner of data)
> > > >> > should keep track of lifecycle of the data in some external
> > > storage/DB.
> > > >> > The owner determines when to delete the data and send the delete
> > > >> request to
> > > >> > kafka. Kafka doesn't know about the content of data but to provide a
> > > >> mean
> > > >> > for deletion.
> > > >> >
> > > >> > 2 , each time compaction runs, it will start from first segments (no
> > > >> > matter if it is compacted or not). The time estimation here is only
> > > used
> > > >> > to determine whether we should run compaction on this log partition.
> > > So
> > > >> we
> > > >> > only need to estimate uncompacted segments.
> > > >> >
> > > >> > On Thu, Aug 16, 2018 at 5:35 PM, Dong Lin 
> > > wrote:
> > > >> >
> > > >> > > Hey Xiongqi,
> > > >> > >
> > > >> > > Thanks for the update. I have two questions for the latest KIP.
> > > >> > >
> > > >> > > 1) The motivation section says that one use case is to delete PII
> > > >> > (Personal
> > > >> > > Identifiable information) data within 7 days while keeping non-PII
> > > >> > > indefinitely in compacted format. I suppose the use-case depends
> > on
> > > >> the
> > > >> > > application to determine when to delete those PII data. Could you
> > > >> explain
> > > >> > > how can application reliably determine the set of keys that should
> > > be
> > > >> > > deleted? Is application required to always messages from the topic
> > > >> after
> > > >> > > every restart and determine the keys to be deleted by looking at
> > > >> message
> > > >> > > timestamp, or is application supposed to persist the key->
> > timstamp
> > > >> > > information in a separate persistent storage system?
> > > >> > >
> > > >> > > 2) It is mentioned in the KIP that "we only need to estimate
> > > earliest
> > > >> > > message timestamp for un-compacted log segments because the
> > deletion
> > > >> > > requests that belong to compacted segments have already been
> > > >> processed".
> > > >> > > Not sure if it is correct. If a segment is compacted before user
> > > sends
> > > >> > > message to delete a key in this segment, it seems that we still
> > need
> > > >> to
> > > >> > > ensure that the segment will be compacted again within the given
> > > time
> > > >> > after
> > > >> > > the deletion is requested, right?
> > > >> > >
> > > >> > > Thanks,
> > > >> > > Dong
> > > >> > >
> > > >> > > On Thu, Aug 16, 2018 at 10:27 AM, xiongqi 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

Brett,

Yes, I will post PR tomorrow.

Xiongqi (Wesley) Wu


On Sun, Sep 2, 2018 at 6:28 PM Brett Rann  wrote:

> +1 (non-binding) from me on the interface. I'd like to see someone familiar
> with
> the code comment on the approach, and note there's a couple of different
> approaches: what's documented in the KIP, and what Xiaohe Dong was working
> on
> here:
>
> https://github.com/dongxiaohe/kafka/tree/dongxiaohe/log-cleaner-compaction-max-lifetime-2.0
>
> If you have code working already Xiongqi Wu could you share a PR? I'd be
> happy
> to start testing.
>
> On Tue, Aug 28, 2018 at 5:57 AM xiongqi wu  wrote:
>
> > Hi All,
> >
> > Do you have any additional comments on this KIP?
> >
> >
> > On Thu, Aug 16, 2018 at 9:17 PM, xiongqi wu  wrote:
> >
> > > on 2)
> > > The offsetmap is built starting from dirty segment.
> > > The compaction starts from the beginning of the log partition. That's
> how
> > > it ensure the deletion of tomb keys.
> > > I will double check tomorrow.
> > >
> > > Xiongqi (Wesley) Wu
> > >
> > >
> > > On Thu, Aug 16, 2018 at 6:46 PM Brett Rann 
> > > wrote:
> > >
> > >> To just clarify a bit on 1. whether there's an external storage/DB
> isn't
> > >> relevant here.
> > >> Compacted topics allow a tombstone record to be sent (a null value
> for a
> > >> key) which
> > >> currently will result in old values for that key being deleted if some
> > >> conditions are met.
> > >> There are existing controls to make sure the old values will stay
> around
> > >> for a minimum
> > >> time at least, but no dedicated control to ensure the tombstone will
> > >> delete
> > >> within a
> > >> maximum time.
> > >>
> > >> One popular reason that maximum time for deletion is desirable right
> now
> > >> is
> > >> GDPR with
> > >> PII. But we're not proposing any GDPR awareness in kafka, just being
> > able
> > >> to guarantee
> > >> a max time where a tombstoned key will be removed from the compacted
> > >> topic.
> > >>
> > >> on 2)
> > >> huh, i thought it kept track of the first dirty segment and didn't
> > >> recompact older "clean" ones.
> > >> But I didn't look at code or test for that.
> > >>
> > >> On Fri, Aug 17, 2018 at 10:57 AM xiongqi wu 
> > wrote:
> > >>
> > >> > 1, Owner of data (in this sense, kafka is the not the owner of data)
> > >> > should keep track of lifecycle of the data in some external
> > storage/DB.
> > >> > The owner determines when to delete the data and send the delete
> > >> request to
> > >> > kafka. Kafka doesn't know about the content of data but to provide a
> > >> mean
> > >> > for deletion.
> > >> >
> > >> > 2 , each time compaction runs, it will start from first segments (no
> > >> > matter if it is compacted or not). The time estimation here is only
> > used
> > >> > to determine whether we should run compaction on this log partition.
> > So
> > >> we
> > >> > only need to estimate uncompacted segments.
> > >> >
> > >> > On Thu, Aug 16, 2018 at 5:35 PM, Dong Lin 
> > wrote:
> > >> >
> > >> > > Hey Xiongqi,
> > >> > >
> > >> > > Thanks for the update. I have two questions for the latest KIP.
> > >> > >
> > >> > > 1) The motivation section says that one use case is to delete PII
> > >> > (Personal
> > >> > > Identifiable information) data within 7 days while keeping non-PII
> > >> > > indefinitely in compacted format. I suppose the use-case depends
> on
> > >> the
> > >> > > application to determine when to delete those PII data. Could you
> > >> explain
> > >> > > how can application reliably determine the set of keys that should
> > be
> > >> > > deleted? Is application required to always messages from the topic
> > >> after
> > >> > > every restart and determine the keys to be deleted by looking at
> > >> message
> > >> > > timestamp, or is application supposed to persist the key->
> timstamp
> > >> > > information in a separate persistent storage system?
> > >> > >
> > >> > > 2) It is mentioned in the KIP that "we only need to estimate
> > earliest
> > >> > > message timestamp for un-compacted log segments because the
> deletion
> > >> > > requests that belong to compacted segments have already been
> > >> processed".
> > >> > > Not sure if it is correct. If a segment is compacted before user
> > sends
> > >> > > message to delete a key in this segment, it seems that we still
> need
> > >> to
> > >> > > ensure that the segment will be compacted again within the given
> > time
> > >> > after
> > >> > > the deletion is requested, right?
> > >> > >
> > >> > > Thanks,
> > >> > > Dong
> > >> > >
> > >> > > On Thu, Aug 16, 2018 at 10:27 AM, xiongqi wu  >
> > >> > wrote:
> > >> > >
> > >> > > > Hi Xiaohe,
> > >> > > >
> > >> > > > Quick note:
> > >> > > > 1) Use minimum of segment.ms and max.compaction.lag.ms
> > >> > > >  > 
> > >> >  > >>
> > >> > > >
> > >> > > > 2) I am not sure if I get your second question. first, we have
> > >> jitter
> > >> > > when
> > 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Brett Rann]

+1 (non-binding) from me on the interface. I'd like to see someone familiar
with
the code comment on the approach, and note there's a couple of different
approaches: what's documented in the KIP, and what Xiaohe Dong was working
on
here:
https://github.com/dongxiaohe/kafka/tree/dongxiaohe/log-cleaner-compaction-max-lifetime-2.0

If you have code working already Xiongqi Wu could you share a PR? I'd be
happy
to start testing.

On Tue, Aug 28, 2018 at 5:57 AM xiongqi wu  wrote:

> Hi All,
>
> Do you have any additional comments on this KIP?
>
>
> On Thu, Aug 16, 2018 at 9:17 PM, xiongqi wu  wrote:
>
> > on 2)
> > The offsetmap is built starting from dirty segment.
> > The compaction starts from the beginning of the log partition. That's how
> > it ensure the deletion of tomb keys.
> > I will double check tomorrow.
> >
> > Xiongqi (Wesley) Wu
> >
> >
> > On Thu, Aug 16, 2018 at 6:46 PM Brett Rann 
> > wrote:
> >
> >> To just clarify a bit on 1. whether there's an external storage/DB isn't
> >> relevant here.
> >> Compacted topics allow a tombstone record to be sent (a null value for a
> >> key) which
> >> currently will result in old values for that key being deleted if some
> >> conditions are met.
> >> There are existing controls to make sure the old values will stay around
> >> for a minimum
> >> time at least, but no dedicated control to ensure the tombstone will
> >> delete
> >> within a
> >> maximum time.
> >>
> >> One popular reason that maximum time for deletion is desirable right now
> >> is
> >> GDPR with
> >> PII. But we're not proposing any GDPR awareness in kafka, just being
> able
> >> to guarantee
> >> a max time where a tombstoned key will be removed from the compacted
> >> topic.
> >>
> >> on 2)
> >> huh, i thought it kept track of the first dirty segment and didn't
> >> recompact older "clean" ones.
> >> But I didn't look at code or test for that.
> >>
> >> On Fri, Aug 17, 2018 at 10:57 AM xiongqi wu 
> wrote:
> >>
> >> > 1, Owner of data (in this sense, kafka is the not the owner of data)
> >> > should keep track of lifecycle of the data in some external
> storage/DB.
> >> > The owner determines when to delete the data and send the delete
> >> request to
> >> > kafka. Kafka doesn't know about the content of data but to provide a
> >> mean
> >> > for deletion.
> >> >
> >> > 2 , each time compaction runs, it will start from first segments (no
> >> > matter if it is compacted or not). The time estimation here is only
> used
> >> > to determine whether we should run compaction on this log partition.
> So
> >> we
> >> > only need to estimate uncompacted segments.
> >> >
> >> > On Thu, Aug 16, 2018 at 5:35 PM, Dong Lin 
> wrote:
> >> >
> >> > > Hey Xiongqi,
> >> > >
> >> > > Thanks for the update. I have two questions for the latest KIP.
> >> > >
> >> > > 1) The motivation section says that one use case is to delete PII
> >> > (Personal
> >> > > Identifiable information) data within 7 days while keeping non-PII
> >> > > indefinitely in compacted format. I suppose the use-case depends on
> >> the
> >> > > application to determine when to delete those PII data. Could you
> >> explain
> >> > > how can application reliably determine the set of keys that should
> be
> >> > > deleted? Is application required to always messages from the topic
> >> after
> >> > > every restart and determine the keys to be deleted by looking at
> >> message
> >> > > timestamp, or is application supposed to persist the key-> timstamp
> >> > > information in a separate persistent storage system?
> >> > >
> >> > > 2) It is mentioned in the KIP that "we only need to estimate
> earliest
> >> > > message timestamp for un-compacted log segments because the deletion
> >> > > requests that belong to compacted segments have already been
> >> processed".
> >> > > Not sure if it is correct. If a segment is compacted before user
> sends
> >> > > message to delete a key in this segment, it seems that we still need
> >> to
> >> > > ensure that the segment will be compacted again within the given
> time
> >> > after
> >> > > the deletion is requested, right?
> >> > >
> >> > > Thanks,
> >> > > Dong
> >> > >
> >> > > On Thu, Aug 16, 2018 at 10:27 AM, xiongqi wu 
> >> > wrote:
> >> > >
> >> > > > Hi Xiaohe,
> >> > > >
> >> > > > Quick note:
> >> > > > 1) Use minimum of segment.ms and max.compaction.lag.ms
> >> > > >  
> >> >  >>
> >> > > >
> >> > > > 2) I am not sure if I get your second question. first, we have
> >> jitter
> >> > > when
> >> > > > we roll the active segment. second, on each compaction, we compact
> >> upto
> >> > > > the offsetmap could allow. Those will not lead to perfect
> compaction
> >> > > storm
> >> > > > overtime. In addition, I expect we are setting
> >> max.compaction.lag.ms
> >> > on
> >> > > > the order of days.
> >> > > >
> >> > > > 3) I don't have access to the confluent community slack for now. I
> >> am

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

Hi All,

Do you have any additional comments on this KIP?


On Thu, Aug 16, 2018 at 9:17 PM, xiongqi wu  wrote:

> on 2)
> The offsetmap is built starting from dirty segment.
> The compaction starts from the beginning of the log partition.  That's how
> it ensure the deletion of tomb keys.
> I will double check tomorrow.
>
> Xiongqi (Wesley) Wu
>
>
> On Thu, Aug 16, 2018 at 6:46 PM Brett Rann 
> wrote:
>
>> To just clarify a bit on 1.  whether there's an external storage/DB isn't
>> relevant here.
>> Compacted topics allow a tombstone record to be sent (a null value for a
>> key) which
>> currently will result in old values for that key being deleted if some
>> conditions are met.
>> There are existing controls to make sure the old values will stay around
>> for a minimum
>> time at least, but no dedicated control to ensure the tombstone will
>> delete
>> within a
>> maximum time.
>>
>> One popular reason that maximum time for deletion is desirable right now
>> is
>> GDPR with
>> PII. But we're not proposing any GDPR awareness in kafka, just being able
>> to guarantee
>> a max time where a tombstoned key will be removed from the compacted
>> topic.
>>
>> on 2)
>> huh, i thought it kept track of the first dirty segment and didn't
>> recompact older "clean" ones.
>> But I didn't look at code or test for that.
>>
>> On Fri, Aug 17, 2018 at 10:57 AM xiongqi wu  wrote:
>>
>> > 1, Owner of data (in this sense, kafka is the not the owner of data)
>> > should keep track of lifecycle of the data in some external storage/DB.
>> > The owner determines when to delete the data and send the delete
>> request to
>> > kafka. Kafka doesn't know about the content of data but to provide a
>> mean
>> > for deletion.
>> >
>> > 2 , each time compaction runs, it will start from first segments (no
>> > matter if it is compacted or not). The time estimation here is only used
>> > to determine whether we should run compaction on this log partition. So
>> we
>> > only need to estimate uncompacted segments.
>> >
>> > On Thu, Aug 16, 2018 at 5:35 PM, Dong Lin  wrote:
>> >
>> > > Hey Xiongqi,
>> > >
>> > > Thanks for the update. I have two questions for the latest KIP.
>> > >
>> > > 1) The motivation section says that one use case is to delete PII
>> > (Personal
>> > > Identifiable information) data within 7 days while keeping non-PII
>> > > indefinitely in compacted format. I suppose the use-case depends on
>> the
>> > > application to determine when to delete those PII data. Could you
>> explain
>> > > how can application reliably determine the set of keys that should be
>> > > deleted? Is application required to always messages from the topic
>> after
>> > > every restart and determine the keys to be deleted by looking at
>> message
>> > > timestamp, or is application supposed to persist the key-> timstamp
>> > > information in a separate persistent storage system?
>> > >
>> > > 2) It is mentioned in the KIP that "we only need to estimate earliest
>> > > message timestamp for un-compacted log segments because the deletion
>> > > requests that belong to compacted segments have already been
>> processed".
>> > > Not sure if it is correct. If a segment is compacted before user sends
>> > > message to delete a key in this segment, it seems that we still need
>> to
>> > > ensure that the segment will be compacted again within the given time
>> > after
>> > > the deletion is requested, right?
>> > >
>> > > Thanks,
>> > > Dong
>> > >
>> > > On Thu, Aug 16, 2018 at 10:27 AM, xiongqi wu 
>> > wrote:
>> > >
>> > > > Hi Xiaohe,
>> > > >
>> > > > Quick note:
>> > > > 1) Use minimum of segment.ms and max.compaction.lag.ms
>> > > > > > >
>> > > >
>> > > > 2) I am not sure if I get your second question. first, we have
>> jitter
>> > > when
>> > > > we roll the active segment. second, on each compaction, we compact
>> upto
>> > > > the offsetmap could allow. Those will not lead to perfect compaction
>> > > storm
>> > > > overtime. In addition, I expect we are setting
>> max.compaction.lag.ms
>> > on
>> > > > the order of days.
>> > > >
>> > > > 3) I don't have access to the confluent community slack for now. I
>> am
>> > > > reachable via the google handle out.
>> > > > To avoid the double effort, here is my plan:
>> > > > a) Collect more feedback and feature requriement on the KIP.
>> > > > b) Wait unitl this KIP is approved.
>> > > > c) I will address any additional requirements in the implementation.
>> > (My
>> > > > current implementation only complies to whatever described in the
>> KIP
>> > > now)
>> > > > d) I can share the code with the you and community see you want to
>> add
>> > > > anything.
>> > > > e) submission through committee
>> > > >
>> > > >
>> > > > On Wed, Aug 15, 2018 at 11:42 PM, XIAOHE DONG <
>> dannyriv...@gmail.com>
>> > > > wrote:
>> > > >
>> > > > > Hi Xiongqi
>> > > > >
>> > > > > Thanks for thinking about implementing this as well. :)
>> > > > >
>> > > > > I was 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

on 2)
The offsetmap is built starting from dirty segment.
The compaction starts from the beginning of the log partition.  That's how
it ensure the deletion of tomb keys.
I will double check tomorrow.

Xiongqi (Wesley) Wu


On Thu, Aug 16, 2018 at 6:46 PM Brett Rann 
wrote:

> To just clarify a bit on 1.  whether there's an external storage/DB isn't
> relevant here.
> Compacted topics allow a tombstone record to be sent (a null value for a
> key) which
> currently will result in old values for that key being deleted if some
> conditions are met.
> There are existing controls to make sure the old values will stay around
> for a minimum
> time at least, but no dedicated control to ensure the tombstone will delete
> within a
> maximum time.
>
> One popular reason that maximum time for deletion is desirable right now is
> GDPR with
> PII. But we're not proposing any GDPR awareness in kafka, just being able
> to guarantee
> a max time where a tombstoned key will be removed from the compacted topic.
>
> on 2)
> huh, i thought it kept track of the first dirty segment and didn't
> recompact older "clean" ones.
> But I didn't look at code or test for that.
>
> On Fri, Aug 17, 2018 at 10:57 AM xiongqi wu  wrote:
>
> > 1, Owner of data (in this sense, kafka is the not the owner of data)
> > should keep track of lifecycle of the data in some external storage/DB.
> > The owner determines when to delete the data and send the delete request
> to
> > kafka. Kafka doesn't know about the content of data but to provide a mean
> > for deletion.
> >
> > 2 , each time compaction runs, it will start from first segments (no
> > matter if it is compacted or not). The time estimation here is only used
> > to determine whether we should run compaction on this log partition. So
> we
> > only need to estimate uncompacted segments.
> >
> > On Thu, Aug 16, 2018 at 5:35 PM, Dong Lin  wrote:
> >
> > > Hey Xiongqi,
> > >
> > > Thanks for the update. I have two questions for the latest KIP.
> > >
> > > 1) The motivation section says that one use case is to delete PII
> > (Personal
> > > Identifiable information) data within 7 days while keeping non-PII
> > > indefinitely in compacted format. I suppose the use-case depends on the
> > > application to determine when to delete those PII data. Could you
> explain
> > > how can application reliably determine the set of keys that should be
> > > deleted? Is application required to always messages from the topic
> after
> > > every restart and determine the keys to be deleted by looking at
> message
> > > timestamp, or is application supposed to persist the key-> timstamp
> > > information in a separate persistent storage system?
> > >
> > > 2) It is mentioned in the KIP that "we only need to estimate earliest
> > > message timestamp for un-compacted log segments because the deletion
> > > requests that belong to compacted segments have already been
> processed".
> > > Not sure if it is correct. If a segment is compacted before user sends
> > > message to delete a key in this segment, it seems that we still need to
> > > ensure that the segment will be compacted again within the given time
> > after
> > > the deletion is requested, right?
> > >
> > > Thanks,
> > > Dong
> > >
> > > On Thu, Aug 16, 2018 at 10:27 AM, xiongqi wu 
> > wrote:
> > >
> > > > Hi Xiaohe,
> > > >
> > > > Quick note:
> > > > 1) Use minimum of segment.ms and max.compaction.lag.ms
> > > >  > >
> > > >
> > > > 2) I am not sure if I get your second question. first, we have jitter
> > > when
> > > > we roll the active segment. second, on each compaction, we compact
> upto
> > > > the offsetmap could allow. Those will not lead to perfect compaction
> > > storm
> > > > overtime. In addition, I expect we are setting max.compaction.lag.ms
> > on
> > > > the order of days.
> > > >
> > > > 3) I don't have access to the confluent community slack for now. I am
> > > > reachable via the google handle out.
> > > > To avoid the double effort, here is my plan:
> > > > a) Collect more feedback and feature requriement on the KIP.
> > > > b) Wait unitl this KIP is approved.
> > > > c) I will address any additional requirements in the implementation.
> > (My
> > > > current implementation only complies to whatever described in the KIP
> > > now)
> > > > d) I can share the code with the you and community see you want to
> add
> > > > anything.
> > > > e) submission through committee
> > > >
> > > >
> > > > On Wed, Aug 15, 2018 at 11:42 PM, XIAOHE DONG  >
> > > > wrote:
> > > >
> > > > > Hi Xiongqi
> > > > >
> > > > > Thanks for thinking about implementing this as well. :)
> > > > >
> > > > > I was thinking about using `segment.ms` to trigger the segment
> roll.
> > > > > Also, its value can be the largest time bias for the record
> deletion.
> > > For
> > > > > example, if the `segment.ms` is 1 day and `max.compaction.ms` is
> 30
> > > > days,
> > > > > the compaction may happen around 31 days.

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Brett Rann]

To just clarify a bit on 1.  whether there's an external storage/DB isn't
relevant here.
Compacted topics allow a tombstone record to be sent (a null value for a
key) which
currently will result in old values for that key being deleted if some
conditions are met.
There are existing controls to make sure the old values will stay around
for a minimum
time at least, but no dedicated control to ensure the tombstone will delete
within a
maximum time.

One popular reason that maximum time for deletion is desirable right now is
GDPR with
PII. But we're not proposing any GDPR awareness in kafka, just being able
to guarantee
a max time where a tombstoned key will be removed from the compacted topic.

on 2)
huh, i thought it kept track of the first dirty segment and didn't
recompact older "clean" ones.
But I didn't look at code or test for that.

On Fri, Aug 17, 2018 at 10:57 AM xiongqi wu  wrote:

> 1, Owner of data (in this sense, kafka is the not the owner of data)
> should keep track of lifecycle of the data in some external storage/DB.
> The owner determines when to delete the data and send the delete request to
> kafka. Kafka doesn't know about the content of data but to provide a mean
> for deletion.
>
> 2 , each time compaction runs, it will start from first segments (no
> matter if it is compacted or not). The time estimation here is only used
> to determine whether we should run compaction on this log partition. So we
> only need to estimate uncompacted segments.
>
> On Thu, Aug 16, 2018 at 5:35 PM, Dong Lin  wrote:
>
> > Hey Xiongqi,
> >
> > Thanks for the update. I have two questions for the latest KIP.
> >
> > 1) The motivation section says that one use case is to delete PII
> (Personal
> > Identifiable information) data within 7 days while keeping non-PII
> > indefinitely in compacted format. I suppose the use-case depends on the
> > application to determine when to delete those PII data. Could you explain
> > how can application reliably determine the set of keys that should be
> > deleted? Is application required to always messages from the topic after
> > every restart and determine the keys to be deleted by looking at message
> > timestamp, or is application supposed to persist the key-> timstamp
> > information in a separate persistent storage system?
> >
> > 2) It is mentioned in the KIP that "we only need to estimate earliest
> > message timestamp for un-compacted log segments because the deletion
> > requests that belong to compacted segments have already been processed".
> > Not sure if it is correct. If a segment is compacted before user sends
> > message to delete a key in this segment, it seems that we still need to
> > ensure that the segment will be compacted again within the given time
> after
> > the deletion is requested, right?
> >
> > Thanks,
> > Dong
> >
> > On Thu, Aug 16, 2018 at 10:27 AM, xiongqi wu 
> wrote:
> >
> > > Hi Xiaohe,
> > >
> > > Quick note:
> > > 1) Use minimum of segment.ms and max.compaction.lag.ms
> > >  >
> > >
> > > 2) I am not sure if I get your second question. first, we have jitter
> > when
> > > we roll the active segment. second, on each compaction, we compact upto
> > > the offsetmap could allow. Those will not lead to perfect compaction
> > storm
> > > overtime. In addition, I expect we are setting max.compaction.lag.ms
> on
> > > the order of days.
> > >
> > > 3) I don't have access to the confluent community slack for now. I am
> > > reachable via the google handle out.
> > > To avoid the double effort, here is my plan:
> > > a) Collect more feedback and feature requriement on the KIP.
> > > b) Wait unitl this KIP is approved.
> > > c) I will address any additional requirements in the implementation.
> (My
> > > current implementation only complies to whatever described in the KIP
> > now)
> > > d) I can share the code with the you and community see you want to add
> > > anything.
> > > e) submission through committee
> > >
> > >
> > > On Wed, Aug 15, 2018 at 11:42 PM, XIAOHE DONG 
> > > wrote:
> > >
> > > > Hi Xiongqi
> > > >
> > > > Thanks for thinking about implementing this as well. :)
> > > >
> > > > I was thinking about using `segment.ms` to trigger the segment roll.
> > > > Also, its value can be the largest time bias for the record deletion.
> > For
> > > > example, if the `segment.ms` is 1 day and `max.compaction.ms` is 30
> > > days,
> > > > the compaction may happen around 31 days.
> > > >
> > > > For my curiosity, is there a way we can do some performance test for
> > this
> > > > and any tools you can recommend. As you know, previously, it is
> cleaned
> > > up
> > > > by respecting dirty ratio, but now it may happen anytime if max lag
> has
> > > > passed for each message. I wonder what would happen if clients send
> > huge
> > > > amount of tombstone records at the same time.
> > > >
> > > > I am looking forward to have a quick chat with you to avoid double
> > effort
> > > > on this. 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

1,  Owner of data (in this sense, kafka is the not the owner of data)
should keep track of lifecycle of the data in some external storage/DB.
The owner determines when to delete the data and send the delete request to
kafka. Kafka doesn't know about the content of data but to provide a mean
for deletion.

2 ,  each time compaction runs, it will start from first segments (no
matter if it is compacted or not).  The time estimation here is only used
to determine whether we should run compaction on this log partition.  So we
only need to estimate uncompacted segments.

On Thu, Aug 16, 2018 at 5:35 PM, Dong Lin  wrote:

> Hey Xiongqi,
>
> Thanks for the update. I have two questions for the latest KIP.
>
> 1) The motivation section says that one use case is to delete PII (Personal
> Identifiable information) data within 7 days while keeping non-PII
> indefinitely in compacted format. I suppose the use-case depends on the
> application to determine when to delete those PII data. Could you explain
> how can application reliably determine the set of keys that should be
> deleted? Is application required to always messages from the topic after
> every restart and determine the keys to be deleted by looking at message
> timestamp, or is application supposed to persist the key-> timstamp
> information in a separate persistent storage system?
>
> 2) It is mentioned in the KIP that "we only need to estimate earliest
> message timestamp for un-compacted log segments because the deletion
> requests that belong to compacted segments have already been processed".
> Not sure if it is correct. If a segment is compacted before user sends
> message to delete a key in this segment, it seems that we still need to
> ensure that the segment will be compacted again within the given time after
> the deletion is requested, right?
>
> Thanks,
> Dong
>
> On Thu, Aug 16, 2018 at 10:27 AM, xiongqi wu  wrote:
>
> > Hi Xiaohe,
> >
> > Quick note:
> > 1) Use minimum of segment.ms and max.compaction.lag.ms
> > 
> >
> > 2) I am not sure if I get your second question.  first, we have jitter
> when
> > we roll the active segment. second, on each  compaction, we compact upto
> > the offsetmap could allow.  Those will not lead to perfect compaction
> storm
> > overtime.  In addition, I expect we are setting max.compaction.lag.ms on
> > the order of days.
> >
> > 3) I don't have access to the confluent community slack for now. I am
> > reachable via the google handle out.
> > To avoid the double effort, here is my plan:
> > a) Collect more feedback and feature requriement on the KIP.
> > b) Wait unitl this KIP is approved.
> > c) I will address any additional requirements in the implementation.  (My
> > current implementation only complies to whatever described in the KIP
> now)
> > d) I can share the code with the you and community see you want to add
> > anything.
> > e) submission through committee
> >
> >
> > On Wed, Aug 15, 2018 at 11:42 PM, XIAOHE DONG 
> > wrote:
> >
> > > Hi Xiongqi
> > >
> > > Thanks for thinking about implementing this as well. :)
> > >
> > > I was thinking about using `segment.ms` to trigger the segment roll.
> > > Also, its value can be the largest time bias for the record deletion.
> For
> > > example, if the `segment.ms` is 1 day and `max.compaction.ms` is 30
> > days,
> > > the compaction may happen around 31 days.
> > >
> > > For my curiosity, is there a way we can do some performance test for
> this
> > > and any tools you can recommend. As you know, previously, it is cleaned
> > up
> > > by respecting dirty ratio, but now it may happen anytime if max lag has
> > > passed for each message. I wonder what would happen if clients send
> huge
> > > amount of tombstone records at the same time.
> > >
> > > I am looking forward to have a quick chat with you to avoid double
> effort
> > > on this. I am in confluent community slack during the work time. My
> name
> > is
> > > Xiaohe Dong. :)
> > >
> > > Rgds
> > > Xiaohe Dong
> > >
> > >
> > >
> > > On 2018/08/16 01:22:22, xiongqi wu  wrote:
> > > > Brett,
> > > >
> > > > Thank you for your comments.
> > > > I was thinking since we already has immediate compaction setting by
> > > setting
> > > > min dirty ratio to 0, so I decide to use "0" as disabled state.
> > > > I am ok to go with -1(disable), 0 (immediate) options.
> > > >
> > > > For the implementation, there are a few differences between mine and
> > > > "Xiaohe Dong"'s :
> > > > 1) I used the estimated creation time of a log segment instead of
> > largest
> > > > timestamp of a log to determine the compaction eligibility, because a
> > log
> > > > segment might stay as an active segment up to "max compaction lag".
> > (see
> > > > the KIP for detail).
> > > > 2) I measure how much bytes that we must clean to follow the "max
> > > > compaction lag" rule, and use that to determine the order of
> > compaction.
> > > > 3) force active segment to roll to follow the "max compaction lag"
> > > 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Dong Lin]

Hey Xiongqi,

Thanks for the update. I have two questions for the latest KIP.

1) The motivation section says that one use case is to delete PII (Personal
Identifiable information) data within 7 days while keeping non-PII
indefinitely in compacted format. I suppose the use-case depends on the
application to determine when to delete those PII data. Could you explain
how can application reliably determine the set of keys that should be
deleted? Is application required to always messages from the topic after
every restart and determine the keys to be deleted by looking at message
timestamp, or is application supposed to persist the key-> timstamp
information in a separate persistent storage system?

2) It is mentioned in the KIP that "we only need to estimate earliest
message timestamp for un-compacted log segments because the deletion
requests that belong to compacted segments have already been processed".
Not sure if it is correct. If a segment is compacted before user sends
message to delete a key in this segment, it seems that we still need to
ensure that the segment will be compacted again within the given time after
the deletion is requested, right?

Thanks,
Dong

On Thu, Aug 16, 2018 at 10:27 AM, xiongqi wu  wrote:

> Hi Xiaohe,
>
> Quick note:
> 1) Use minimum of segment.ms and max.compaction.lag.ms
> 
>
> 2) I am not sure if I get your second question.  first, we have jitter when
> we roll the active segment. second, on each  compaction, we compact upto
> the offsetmap could allow.  Those will not lead to perfect compaction storm
> overtime.  In addition, I expect we are setting max.compaction.lag.ms on
> the order of days.
>
> 3) I don't have access to the confluent community slack for now. I am
> reachable via the google handle out.
> To avoid the double effort, here is my plan:
> a) Collect more feedback and feature requriement on the KIP.
> b) Wait unitl this KIP is approved.
> c) I will address any additional requirements in the implementation.  (My
> current implementation only complies to whatever described in the KIP now)
> d) I can share the code with the you and community see you want to add
> anything.
> e) submission through committee
>
>
> On Wed, Aug 15, 2018 at 11:42 PM, XIAOHE DONG 
> wrote:
>
> > Hi Xiongqi
> >
> > Thanks for thinking about implementing this as well. :)
> >
> > I was thinking about using `segment.ms` to trigger the segment roll.
> > Also, its value can be the largest time bias for the record deletion. For
> > example, if the `segment.ms` is 1 day and `max.compaction.ms` is 30
> days,
> > the compaction may happen around 31 days.
> >
> > For my curiosity, is there a way we can do some performance test for this
> > and any tools you can recommend. As you know, previously, it is cleaned
> up
> > by respecting dirty ratio, but now it may happen anytime if max lag has
> > passed for each message. I wonder what would happen if clients send huge
> > amount of tombstone records at the same time.
> >
> > I am looking forward to have a quick chat with you to avoid double effort
> > on this. I am in confluent community slack during the work time. My name
> is
> > Xiaohe Dong. :)
> >
> > Rgds
> > Xiaohe Dong
> >
> >
> >
> > On 2018/08/16 01:22:22, xiongqi wu  wrote:
> > > Brett,
> > >
> > > Thank you for your comments.
> > > I was thinking since we already has immediate compaction setting by
> > setting
> > > min dirty ratio to 0, so I decide to use "0" as disabled state.
> > > I am ok to go with -1(disable), 0 (immediate) options.
> > >
> > > For the implementation, there are a few differences between mine and
> > > "Xiaohe Dong"'s :
> > > 1) I used the estimated creation time of a log segment instead of
> largest
> > > timestamp of a log to determine the compaction eligibility, because a
> log
> > > segment might stay as an active segment up to "max compaction lag".
> (see
> > > the KIP for detail).
> > > 2) I measure how much bytes that we must clean to follow the "max
> > > compaction lag" rule, and use that to determine the order of
> compaction.
> > > 3) force active segment to roll to follow the "max compaction lag"
> > >
> > > I can share my code so we can coordinate.
> > >
> > > I haven't think about a new API to force a compaction. what is the use
> > case
> > > for this one?
> > >
> > >
> > > On Wed, Aug 15, 2018 at 5:33 PM, Brett Rann  >
> > > wrote:
> > >
> > > > We've been looking into this too.
> > > >
> > > > Mailing list:
> > > > https://lists.apache.org/thread.html/ed7f6a6589f94e8c2a705553f364ef
> > > > 599cb6915e4c3ba9b561e610e4@%3Cdev.kafka.apache.org%3E
> > > > jira wish: https://issues.apache.org/jira/browse/KAFKA-7137
> > > > confluent slack discussion:
> > > > https://confluentcommunity.slack.com/archives/C49R61XMM/
> > p153076012139
> > > >
> > > > A person on my team has started on code so you might want to
> > coordinate:
> > > > https://github.com/dongxiaohe/kafka/tree/dongxiaohe/log-
> > > > 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

Hi Xiaohe,

Quick note:
1) Use minimum of segment.ms and max.compaction.lag.ms


2) I am not sure if I get your second question.  first, we have jitter when
we roll the active segment. second, on each  compaction, we compact upto
the offsetmap could allow.  Those will not lead to perfect compaction storm
overtime.  In addition, I expect we are setting max.compaction.lag.ms on
the order of days.

3) I don't have access to the confluent community slack for now. I am
reachable via the google handle out.
To avoid the double effort, here is my plan:
a) Collect more feedback and feature requriement on the KIP.
b) Wait unitl this KIP is approved.
c) I will address any additional requirements in the implementation.  (My
current implementation only complies to whatever described in the KIP now)
d) I can share the code with the you and community see you want to add
anything.
e) submission through committee


On Wed, Aug 15, 2018 at 11:42 PM, XIAOHE DONG  wrote:

> Hi Xiongqi
>
> Thanks for thinking about implementing this as well. :)
>
> I was thinking about using `segment.ms` to trigger the segment roll.
> Also, its value can be the largest time bias for the record deletion. For
> example, if the `segment.ms` is 1 day and `max.compaction.ms` is 30 days,
> the compaction may happen around 31 days.
>
> For my curiosity, is there a way we can do some performance test for this
> and any tools you can recommend. As you know, previously, it is cleaned up
> by respecting dirty ratio, but now it may happen anytime if max lag has
> passed for each message. I wonder what would happen if clients send huge
> amount of tombstone records at the same time.
>
> I am looking forward to have a quick chat with you to avoid double effort
> on this. I am in confluent community slack during the work time. My name is
> Xiaohe Dong. :)
>
> Rgds
> Xiaohe Dong
>
>
>
> On 2018/08/16 01:22:22, xiongqi wu  wrote:
> > Brett,
> >
> > Thank you for your comments.
> > I was thinking since we already has immediate compaction setting by
> setting
> > min dirty ratio to 0, so I decide to use "0" as disabled state.
> > I am ok to go with -1(disable), 0 (immediate) options.
> >
> > For the implementation, there are a few differences between mine and
> > "Xiaohe Dong"'s :
> > 1) I used the estimated creation time of a log segment instead of largest
> > timestamp of a log to determine the compaction eligibility, because a log
> > segment might stay as an active segment up to "max compaction lag". (see
> > the KIP for detail).
> > 2) I measure how much bytes that we must clean to follow the "max
> > compaction lag" rule, and use that to determine the order of compaction.
> > 3) force active segment to roll to follow the "max compaction lag"
> >
> > I can share my code so we can coordinate.
> >
> > I haven't think about a new API to force a compaction. what is the use
> case
> > for this one?
> >
> >
> > On Wed, Aug 15, 2018 at 5:33 PM, Brett Rann 
> > wrote:
> >
> > > We've been looking into this too.
> > >
> > > Mailing list:
> > > https://lists.apache.org/thread.html/ed7f6a6589f94e8c2a705553f364ef
> > > 599cb6915e4c3ba9b561e610e4@%3Cdev.kafka.apache.org%3E
> > > jira wish: https://issues.apache.org/jira/browse/KAFKA-7137
> > > confluent slack discussion:
> > > https://confluentcommunity.slack.com/archives/C49R61XMM/
> p153076012139
> > >
> > > A person on my team has started on code so you might want to
> coordinate:
> > > https://github.com/dongxiaohe/kafka/tree/dongxiaohe/log-
> > > cleaner-compaction-max-lifetime-2.0
> > >
> > >  He's been working with Jason Gustafson and James Chen around the
> changes.
> > > You can ping him on confluent slack as Xiaohe Dong.
> > >
> > > It's great to know others are thinking on it as well.
> > >
> > > You've added the requirement to force a segment roll which we hadn't
> gotten
> > > to yet, which is great. I was content with it not including the active
> > > segment.
> > >
> > > > Adding topic level configuration "max.compaction.lag.ms",  and
> > > corresponding broker configuration "log.cleaner.max.compaction.lag.ms
> ",
> > > which is set to 0 (disabled) by default.
> > >
> > > Glancing at some other settings convention seems to me to be -1 for
> > > disabled (or infinite, which is more meaningful here).  0 to me implies
> > > instant, a little quicker than 1.
> > >
> > > We've been trying to think about a way to trigger compaction as well
> > > through an API call, which would need to be flagged somewhere (ZK
> admin/
> > > space?) but we're struggling to think how that would be coordinated
> across
> > > brokers and partitions.  Have you given any thought to that?
> > >
> > >
> > >
> > >
> > >
> > >
> > > On Thu, Aug 16, 2018 at 8:44 AM xiongqi wu 
> wrote:
> > >
> > > > Eno, Dong,
> > > >
> > > > I have updated the KIP. We decide not to address the issue that we
> might
> > > > have for both compaction and time retention enabled topics (see the
> > > > rejected alternative item 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

Hi Brett,

In my opinion, we don't need very accurate estimation of first record
timestamp because the compaction processing itself might take
some time. There is no strictly guaranteed that after the time interval
"T", the compaction is finished.  Rather than,
after time "T" + some compaction job interval,  the compaction will start
processing.
For this reason,  using the largest timestamp of previous segment to
estimate creation time of next segments looks sufficient to me.

One major principal is to not to introduce a lot IOs to those partitions
that doesn't require to be compacted.
I want to avoid looking into records or timestamp as much as possible.

Partition movement can alter the segment modification time.  However I see
this as the data is in a transit state.
After another time interval T when data is stabilized in new location, the
compaction will be launched.
We also use largestTimestamp in the time based retention.

On Wed, Aug 15, 2018 at 9:08 PM, Brett Rann 
wrote:

> > (segment.largestTimestamp is lastModified time of the log segment or max
> timestamp we see for the log segment. Due to the lack of record timestamp,
> segment.largestTimestamp might be earlier than the actual timestamp of
> latest record of that segment.).
>
> I'm curious about the mention of last modified time of the segment.  As
> noted back in here
> https://cwiki-test.apache.org/confluence/display/KAFKA/KIP-
> 33+-+Add+a+time+based+log+index
> using the creation/modified time of files is unreliable in cases of
> partitions being moved. It's why all those .timeindex files for each
> partition appeared in 0.10.* I wonder if there's a better way using that to
> get at a timestamp?
>
> On Thu, Aug 16, 2018 at 11:30 AM Brett Rann  wrote:
>
> > An API was suggested by Gwen and James when I discussed it with them. For
> > me I can think of it as a use case for scheduling compaction rather than
> > relying on an config based time trigger.  We're looking at creating some
> > potentially very large compacted topics for event sourcing and from an
> > operators perspective I'm concerned about potential impact of long
> running
> > compaction, especially if multiple topics run back to back.  Having the
> > ability to schedule them at my own time gives some peace of mind for that
> > concern.  Another use case might be a more urgent delete. That could be
> > done manually now by just modifying the max time and waiting for
> compaction
> > to run. But hitting an API end point is a bit cleaner.
> >
> > But in thinking about what that mechanism would be it started to feel
> like
> > it would be a complicated implementation so we've put it aside for now.
> But
> > maybe we just haven't seen the clean way yet.
> >
> >
> >
> > On Thu, Aug 16, 2018 at 11:22 AM xiongqi wu  wrote:
> >
> >> Brett,
> >>
> >> Thank you for your comments.
> >> I was thinking since we already has immediate compaction setting by
> >> setting
> >> min dirty ratio to 0, so I decide to use "0" as disabled state.
> >> I am ok to go with -1(disable), 0 (immediate) options.
> >>
> >> For the implementation, there are a few differences between mine and
> >> "Xiaohe Dong"'s :
> >> 1) I used the estimated creation time of a log segment instead of
> largest
> >> timestamp of a log to determine the compaction eligibility, because a
> log
> >> segment might stay as an active segment up to "max compaction lag". (see
> >> the KIP for detail).
> >> 2) I measure how much bytes that we must clean to follow the "max
> >> compaction lag" rule, and use that to determine the order of compaction.
> >> 3) force active segment to roll to follow the "max compaction lag"
> >>
> >> I can share my code so we can coordinate.
> >>
> >> I haven't think about a new API to force a compaction. what is the use
> >> case
> >> for this one?
> >>
> >>
> >> On Wed, Aug 15, 2018 at 5:33 PM, Brett Rann 
> >> wrote:
> >>
> >> > We've been looking into this too.
> >> >
> >> > Mailing list:
> >> > https://lists.apache.org/thread.html/ed7f6a6589f94e8c2a705553f364ef
> >> 
> >> > 599cb6915e4c3ba9b561e610e4@%3Cdev.kafka.apache.org%3E
> >> > jira wish: https://issues.apache.org/jira/browse/KAFKA-7137
> >> 
> >> > confluent slack discussion:
> >> >
> >> https://confluentcommunity.slack.com/archives/C49R61XMM/
> p153076012139
> >>  p153076012139>
> >> >
> >> > A person on my team has started on code so you might want to
> coordinate:
> >> > https://github.com/dongxiaohe/kafka/tree/dongxiaohe/log-
> >> 
> >> > cleaner-compaction-max-lifetime-2.0
> >> >
> >> > He's been working with Jason Gustafson and James Chen around the
> >> changes.
> >> > You can ping him on confluent slack as Xiaohe Dong.
> >> >
> >> > It's great to know others are thinking on it as well.
> >> >
> >> > 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[XIAOHE DONG]

Hi Xiongqi

Thanks for thinking about implementing this as well. :)

I was thinking about using `segment.ms` to trigger the segment roll. Also, its 
value can be the largest time bias for the record deletion. For example, if the 
`segment.ms` is 1 day and `max.compaction.ms` is 30 days, the compaction may 
happen around 31 days. 

For my curiosity, is there a way we can do some performance test for this and 
any tools you can recommend. As you know, previously, it is cleaned up by 
respecting dirty ratio, but now it may happen anytime if max lag has passed for 
each message. I wonder what would happen if clients send huge amount of 
tombstone records at the same time. 

I am looking forward to have a quick chat with you to avoid double effort on 
this. I am in confluent community slack during the work time. My name is Xiaohe 
Dong. :)

Rgds
Xiaohe Dong



On 2018/08/16 01:22:22, xiongqi wu  wrote: 
> Brett,
> 
> Thank you for your comments.
> I was thinking since we already has immediate compaction setting by setting
> min dirty ratio to 0, so I decide to use "0" as disabled state.
> I am ok to go with -1(disable), 0 (immediate) options.
> 
> For the implementation, there are a few differences between mine and
> "Xiaohe Dong"'s :
> 1) I used the estimated creation time of a log segment instead of largest
> timestamp of a log to determine the compaction eligibility, because a log
> segment might stay as an active segment up to "max compaction lag". (see
> the KIP for detail).
> 2) I measure how much bytes that we must clean to follow the "max
> compaction lag" rule, and use that to determine the order of compaction.
> 3) force active segment to roll to follow the "max compaction lag"
> 
> I can share my code so we can coordinate.
> 
> I haven't think about a new API to force a compaction. what is the use case
> for this one?
> 
> 
> On Wed, Aug 15, 2018 at 5:33 PM, Brett Rann 
> wrote:
> 
> > We've been looking into this too.
> >
> > Mailing list:
> > https://lists.apache.org/thread.html/ed7f6a6589f94e8c2a705553f364ef
> > 599cb6915e4c3ba9b561e610e4@%3Cdev.kafka.apache.org%3E
> > jira wish: https://issues.apache.org/jira/browse/KAFKA-7137
> > confluent slack discussion:
> > https://confluentcommunity.slack.com/archives/C49R61XMM/p153076012139
> >
> > A person on my team has started on code so you might want to coordinate:
> > https://github.com/dongxiaohe/kafka/tree/dongxiaohe/log-
> > cleaner-compaction-max-lifetime-2.0
> >
> >  He's been working with Jason Gustafson and James Chen around the changes.
> > You can ping him on confluent slack as Xiaohe Dong.
> >
> > It's great to know others are thinking on it as well.
> >
> > You've added the requirement to force a segment roll which we hadn't gotten
> > to yet, which is great. I was content with it not including the active
> > segment.
> >
> > > Adding topic level configuration "max.compaction.lag.ms",  and
> > corresponding broker configuration "log.cleaner.max.compaction.lag.ms",
> > which is set to 0 (disabled) by default.
> >
> > Glancing at some other settings convention seems to me to be -1 for
> > disabled (or infinite, which is more meaningful here).  0 to me implies
> > instant, a little quicker than 1.
> >
> > We've been trying to think about a way to trigger compaction as well
> > through an API call, which would need to be flagged somewhere (ZK admin/
> > space?) but we're struggling to think how that would be coordinated across
> > brokers and partitions.  Have you given any thought to that?
> >
> >
> >
> >
> >
> >
> > On Thu, Aug 16, 2018 at 8:44 AM xiongqi wu  wrote:
> >
> > > Eno, Dong,
> > >
> > > I have updated the KIP. We decide not to address the issue that we might
> > > have for both compaction and time retention enabled topics (see the
> > > rejected alternative item 2). This KIP will only ensure log can be
> > > compacted after a specified time-interval.
> > >
> > > As suggested by Dong, we will also enforce "max.compaction.lag.ms" is
> > not
> > > less than "min.compaction.lag.ms".
> > >
> > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-354: Time-based
> > log
> > > compaction policy
> > >  > log compaction policy>
> > >
> > >
> > > On Tue, Aug 14, 2018 at 5:01 PM, xiongqi wu  wrote:
> > >
> > > >
> > > > Per discussion with Dong, he made a very good point that if compaction
> > > > and time based retention are both enabled on a topic, the compaction
> > > might
> > > > prevent records from being deleted on time. The reason is when
> > compacting
> > > > multiple segments into one single segment, the newly created segment
> > will
> > > > have same lastmodified timestamp as latest original segment. We lose
> > the
> > > > timestamp of all original segments except the last one. As a result,
> > > > records might not be deleted as it should be through time based
> > > retention.
> > > >
> > > > With the current KIP proposal, if we want to ensure 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Brett Rann]

> (segment.largestTimestamp is lastModified time of the log segment or max
timestamp we see for the log segment. Due to the lack of record timestamp,
segment.largestTimestamp might be earlier than the actual timestamp of
latest record of that segment.).

I'm curious about the mention of last modified time of the segment.  As
noted back in here
https://cwiki-test.apache.org/confluence/display/KAFKA/KIP-33+-+Add+a+time+based+log+index
using the creation/modified time of files is unreliable in cases of
partitions being moved. It's why all those .timeindex files for each
partition appeared in 0.10.* I wonder if there's a better way using that to
get at a timestamp?

On Thu, Aug 16, 2018 at 11:30 AM Brett Rann  wrote:

> An API was suggested by Gwen and James when I discussed it with them. For
> me I can think of it as a use case for scheduling compaction rather than
> relying on an config based time trigger.  We're looking at creating some
> potentially very large compacted topics for event sourcing and from an
> operators perspective I'm concerned about potential impact of long running
> compaction, especially if multiple topics run back to back.  Having the
> ability to schedule them at my own time gives some peace of mind for that
> concern.  Another use case might be a more urgent delete. That could be
> done manually now by just modifying the max time and waiting for compaction
> to run. But hitting an API end point is a bit cleaner.
>
> But in thinking about what that mechanism would be it started to feel like
> it would be a complicated implementation so we've put it aside for now. But
> maybe we just haven't seen the clean way yet.
>
>
>
> On Thu, Aug 16, 2018 at 11:22 AM xiongqi wu  wrote:
>
>> Brett,
>>
>> Thank you for your comments.
>> I was thinking since we already has immediate compaction setting by
>> setting
>> min dirty ratio to 0, so I decide to use "0" as disabled state.
>> I am ok to go with -1(disable), 0 (immediate) options.
>>
>> For the implementation, there are a few differences between mine and
>> "Xiaohe Dong"'s :
>> 1) I used the estimated creation time of a log segment instead of largest
>> timestamp of a log to determine the compaction eligibility, because a log
>> segment might stay as an active segment up to "max compaction lag". (see
>> the KIP for detail).
>> 2) I measure how much bytes that we must clean to follow the "max
>> compaction lag" rule, and use that to determine the order of compaction.
>> 3) force active segment to roll to follow the "max compaction lag"
>>
>> I can share my code so we can coordinate.
>>
>> I haven't think about a new API to force a compaction. what is the use
>> case
>> for this one?
>>
>>
>> On Wed, Aug 15, 2018 at 5:33 PM, Brett Rann 
>> wrote:
>>
>> > We've been looking into this too.
>> >
>> > Mailing list:
>> > https://lists.apache.org/thread.html/ed7f6a6589f94e8c2a705553f364ef
>> 
>> > 599cb6915e4c3ba9b561e610e4@%3Cdev.kafka.apache.org%3E
>> > jira wish: https://issues.apache.org/jira/browse/KAFKA-7137
>> 
>> > confluent slack discussion:
>> >
>> https://confluentcommunity.slack.com/archives/C49R61XMM/p153076012139
>> 
>> >
>> > A person on my team has started on code so you might want to coordinate:
>> > https://github.com/dongxiaohe/kafka/tree/dongxiaohe/log-
>> 
>> > cleaner-compaction-max-lifetime-2.0
>> >
>> > He's been working with Jason Gustafson and James Chen around the
>> changes.
>> > You can ping him on confluent slack as Xiaohe Dong.
>> >
>> > It's great to know others are thinking on it as well.
>> >
>> > You've added the requirement to force a segment roll which we hadn't
>> gotten
>> > to yet, which is great. I was content with it not including the active
>> > segment.
>> >
>> > > Adding topic level configuration "max.compaction.lag.ms", and
>> > corresponding broker configuration "log.cleaner.max.compaction.lag.ms",
>> > which is set to 0 (disabled) by default.
>> >
>> > Glancing at some other settings convention seems to me to be -1 for
>> > disabled (or infinite, which is more meaningful here). 0 to me implies
>> > instant, a little quicker than 1.
>> >
>> > We've been trying to think about a way to trigger compaction as well
>> > through an API call, which would need to be flagged somewhere (ZK admin/
>> > space?) but we're struggling to think how that would be coordinated
>> across
>> > brokers and partitions. Have you given any thought to that?
>> >
>> >
>> >
>> >
>> >
>> >
>> > On Thu, Aug 16, 2018 at 8:44 AM xiongqi wu  wrote:
>> >
>> > > Eno, Dong,
>> > >
>> > > I have updated the KIP. We decide not to address the issue that we
>> might
>> > > have for both compaction and time retention enabled topics (see the
>> > > rejected alternative item 2). This KIP will 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Brett Rann]

An API was suggested by Gwen and James when I discussed it with them. For
me I can think of it as a use case for scheduling compaction rather than
relying on an config based time trigger.  We're looking at creating some
potentially very large compacted topics for event sourcing and from an
operators perspective I'm concerned about potential impact of long running
compaction, especially if multiple topics run back to back.  Having the
ability to schedule them at my own time gives some peace of mind for that
concern.  Another use case might be a more urgent delete. That could be
done manually now by just modifying the max time and waiting for compaction
to run. But hitting an API end point is a bit cleaner.

But in thinking about what that mechanism would be it started to feel like
it would be a complicated implementation so we've put it aside for now. But
maybe we just haven't seen the clean way yet.



On Thu, Aug 16, 2018 at 11:22 AM xiongqi wu  wrote:

> Brett,
>
> Thank you for your comments.
> I was thinking since we already has immediate compaction setting by setting
> min dirty ratio to 0, so I decide to use "0" as disabled state.
> I am ok to go with -1(disable), 0 (immediate) options.
>
> For the implementation, there are a few differences between mine and
> "Xiaohe Dong"'s :
> 1) I used the estimated creation time of a log segment instead of largest
> timestamp of a log to determine the compaction eligibility, because a log
> segment might stay as an active segment up to "max compaction lag". (see
> the KIP for detail).
> 2) I measure how much bytes that we must clean to follow the "max
> compaction lag" rule, and use that to determine the order of compaction.
> 3) force active segment to roll to follow the "max compaction lag"
>
> I can share my code so we can coordinate.
>
> I haven't think about a new API to force a compaction. what is the use case
> for this one?
>
>
> On Wed, Aug 15, 2018 at 5:33 PM, Brett Rann 
> wrote:
>
> > We've been looking into this too.
> >
> > Mailing list:
> > https://lists.apache.org/thread.html/ed7f6a6589f94e8c2a705553f364ef
> 
> > 599cb6915e4c3ba9b561e610e4@%3Cdev.kafka.apache.org%3E
> > jira wish: https://issues.apache.org/jira/browse/KAFKA-7137
> 
> > confluent slack discussion:
> >
> https://confluentcommunity.slack.com/archives/C49R61XMM/p153076012139
> 
> >
> > A person on my team has started on code so you might want to coordinate:
> > https://github.com/dongxiaohe/kafka/tree/dongxiaohe/log-
> 
> > cleaner-compaction-max-lifetime-2.0
> >
> > He's been working with Jason Gustafson and James Chen around the changes.
> > You can ping him on confluent slack as Xiaohe Dong.
> >
> > It's great to know others are thinking on it as well.
> >
> > You've added the requirement to force a segment roll which we hadn't
> gotten
> > to yet, which is great. I was content with it not including the active
> > segment.
> >
> > > Adding topic level configuration "max.compaction.lag.ms", and
> > corresponding broker configuration "log.cleaner.max.compaction.lag.ms",
> > which is set to 0 (disabled) by default.
> >
> > Glancing at some other settings convention seems to me to be -1 for
> > disabled (or infinite, which is more meaningful here). 0 to me implies
> > instant, a little quicker than 1.
> >
> > We've been trying to think about a way to trigger compaction as well
> > through an API call, which would need to be flagged somewhere (ZK admin/
> > space?) but we're struggling to think how that would be coordinated
> across
> > brokers and partitions. Have you given any thought to that?
> >
> >
> >
> >
> >
> >
> > On Thu, Aug 16, 2018 at 8:44 AM xiongqi wu  wrote:
> >
> > > Eno, Dong,
> > >
> > > I have updated the KIP. We decide not to address the issue that we
> might
> > > have for both compaction and time retention enabled topics (see the
> > > rejected alternative item 2). This KIP will only ensure log can be
> > > compacted after a specified time-interval.
> > >
> > > As suggested by Dong, we will also enforce "max.compaction.lag.ms" is
> > not
> > > less than "min.compaction.lag.ms".
> > >
> > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-354
>  Time-based
> > log
> > > compaction policy
> > >   Time-based
> > log compaction policy>
> > >
> > >
> > > On Tue, Aug 14, 2018 at 5:01 PM, xiongqi wu 
> wrote:
> > >
> > > >
> > > > Per discussion with Dong, he made a very good point that if
> compaction
> > > > and time based retention are both enabled on a topic, the compaction
> > > might
> > > > prevent records from being deleted on 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

Brett,

Thank you for your comments.
I was thinking since we already has immediate compaction setting by setting
min dirty ratio to 0, so I decide to use "0" as disabled state.
I am ok to go with -1(disable), 0 (immediate) options.

For the implementation, there are a few differences between mine and
"Xiaohe Dong"'s :
1) I used the estimated creation time of a log segment instead of largest
timestamp of a log to determine the compaction eligibility, because a log
segment might stay as an active segment up to "max compaction lag". (see
the KIP for detail).
2) I measure how much bytes that we must clean to follow the "max
compaction lag" rule, and use that to determine the order of compaction.
3) force active segment to roll to follow the "max compaction lag"

I can share my code so we can coordinate.

I haven't think about a new API to force a compaction. what is the use case
for this one?


On Wed, Aug 15, 2018 at 5:33 PM, Brett Rann 
wrote:

> We've been looking into this too.
>
> Mailing list:
> https://lists.apache.org/thread.html/ed7f6a6589f94e8c2a705553f364ef
> 599cb6915e4c3ba9b561e610e4@%3Cdev.kafka.apache.org%3E
> jira wish: https://issues.apache.org/jira/browse/KAFKA-7137
> confluent slack discussion:
> https://confluentcommunity.slack.com/archives/C49R61XMM/p153076012139
>
> A person on my team has started on code so you might want to coordinate:
> https://github.com/dongxiaohe/kafka/tree/dongxiaohe/log-
> cleaner-compaction-max-lifetime-2.0
>
>  He's been working with Jason Gustafson and James Chen around the changes.
> You can ping him on confluent slack as Xiaohe Dong.
>
> It's great to know others are thinking on it as well.
>
> You've added the requirement to force a segment roll which we hadn't gotten
> to yet, which is great. I was content with it not including the active
> segment.
>
> > Adding topic level configuration "max.compaction.lag.ms",  and
> corresponding broker configuration "log.cleaner.max.compaction.lag.ms",
> which is set to 0 (disabled) by default.
>
> Glancing at some other settings convention seems to me to be -1 for
> disabled (or infinite, which is more meaningful here).  0 to me implies
> instant, a little quicker than 1.
>
> We've been trying to think about a way to trigger compaction as well
> through an API call, which would need to be flagged somewhere (ZK admin/
> space?) but we're struggling to think how that would be coordinated across
> brokers and partitions.  Have you given any thought to that?
>
>
>
>
>
>
> On Thu, Aug 16, 2018 at 8:44 AM xiongqi wu  wrote:
>
> > Eno, Dong,
> >
> > I have updated the KIP. We decide not to address the issue that we might
> > have for both compaction and time retention enabled topics (see the
> > rejected alternative item 2). This KIP will only ensure log can be
> > compacted after a specified time-interval.
> >
> > As suggested by Dong, we will also enforce "max.compaction.lag.ms" is
> not
> > less than "min.compaction.lag.ms".
> >
> > https://cwiki.apache.org/confluence/display/KAFKA/KIP-354: Time-based
> log
> > compaction policy
> >  log compaction policy>
> >
> >
> > On Tue, Aug 14, 2018 at 5:01 PM, xiongqi wu  wrote:
> >
> > >
> > > Per discussion with Dong, he made a very good point that if compaction
> > > and time based retention are both enabled on a topic, the compaction
> > might
> > > prevent records from being deleted on time. The reason is when
> compacting
> > > multiple segments into one single segment, the newly created segment
> will
> > > have same lastmodified timestamp as latest original segment. We lose
> the
> > > timestamp of all original segments except the last one. As a result,
> > > records might not be deleted as it should be through time based
> > retention.
> > >
> > > With the current KIP proposal, if we want to ensure timely deletion, we
> > > have the following configurations:
> > > 1) enable time based log compaction only : deletion is done though
> > > overriding the same key
> > > 2) enable time based log retention only: deletion is done though
> > > time-based retention
> > > 3) enable both log compaction and time based retention: Deletion is not
> > > guaranteed.
> > >
> > > Not sure if we have use case 3 and also want deletion to happen on
> time.
> > > There are several options to address deletion issue when enable both
> > > compaction and retention:
> > > A) During log compaction, looking into record timestamp to delete
> expired
> > > records. This can be done in compaction logic itself or use
> > > AdminClient.deleteRecords() . But this assumes we have record
> timestamp.
> > > B) retain the lastModifed time of original segments during log
> > compaction.
> > > This requires extra meta data to record the information or not grouping
> > > multiple segments into one during compaction.
> > >
> > > If we have use case 3 in general, I would prefer solution A and rely on
> > > record timestamp.
> > >
> > >
> > > 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Brett Rann]

Eno,

For us as well the requirement is around compacted topics because they are
the topics that already facilitate selective deletes. Currently they allow
specifying a minimum life time, but lacks the ability to specify a maximum
life time.

For non compacted topics there's no ability to delete individual messages,
they're immutable logs. We treat those with hard rules: Max retention time
on the topic; accept the topic may get truncated; or to not store
information that may be subject to GDPR. (and i've read others use tricks
with encryption and forgetting the decryption key).

Enhancing compaction to support a max compaction time makes the compacted
topics more useful, especially in that it allows the dirty ratio to be used
for its intended purpose while allowing automatic cleaning based on a new
time config.

On Tue, Aug 14, 2018 at 9:00 PM Eno Thereska  wrote:

> Adding to this, what about topics that are not log compacted? As Dong says,
> "one of the GDPR requirement is that we can not keep messages longer than
> e.g. 30 days in storage (e.g. Kafka)". The GDPR requirement must hold
> irrespective of the low level details, on whether the topic is compacted or
> not, right?
>
> Thanks
> Eno
>
>
> On Mon, Aug 13, 2018 at 6:58 PM, Dong Lin  wrote:
>
> > Hey Xiongqi,
> >
> > Thanks for the KIP. I have two questions regarding the use-case for
> meeting
> > GDPR requirement.
> >
> > 1) If I recall correctly, one of the GDPR requirement is that we can not
> > keep messages longer than e.g. 30 days in storage (e.g. Kafka). Say there
> > exists a partition p0 which contains message1 with key1 and message2 with
> > key2. And then user keeps producing messages with key=key2 to this
> > partition. Since message1 with key1 is never overridden, sooner or later
> we
> > will want to delete message1 and keep the latest message with key=key2.
> But
> > currently it looks like log compact logic in Kafka will always put these
> > messages in the same segment. Will this be an issue?
> >
> > 2) The current KIP intends to provide the capability to delete a given
> > message in log compacted topic. Does such use-case also require Kafka to
> > keep the messages produced before the given message? If yes, then we can
> > probably just use AdminClient.deleteRecords() or time-based log retention
> > to meet the use-case requirement. If no, do you know what is the GDPR's
> > requirement on time-to-deletion after user explicitly requests the
> deletion
> > (e.g. 1 hour, 1 day, 7 day)?
> >
> > Thanks,
> > Dong
> >
> >
> > On Mon, Aug 13, 2018 at 3:44 PM, xiongqi wu  wrote:
> >
> > > Hi Eno,
> > >
> > > The GDPR request we are getting here at linkedin is if we get a request
> > to
> > > delete a record through a null key on a log compacted topic,
> > > we want to delete the record via compaction in a given time period
> like 2
> > > days (whatever is required by the policy).
> > >
> > > There might be other issues (such as orphan log segments under certain
> > > conditions) that lead to GDPR problem but they are more like something
> > we
> > > need to fix anyway regardless of GDPR.
> > >
> > >
> > > -- Xiongqi (Wesley) Wu
> > >
> > > On Mon, Aug 13, 2018 at 2:56 PM, Eno Thereska 
> > > wrote:
> > >
> > > > Hello,
> > > >
> > > > Thanks for the KIP. I'd like to see a more precise definition of what
> > > part
> > > > of GDPR you are targeting as well as some sort of verification that
> > this
> > > > KIP actually addresses the problem. Right now I find this a bit
> vague:
> > > >
> > > > "Ability to delete a log message through compaction in a timely
> manner
> > > has
> > > > become an important requirement in some use cases (e.g., GDPR)"
> > > >
> > > >
> > > > Is there any guarantee that after this KIP the GDPR problem is solved
> > or
> > > do
> > > > we need to do something else as well, e.g., more KIPs?
> > > >
> > > >
> > > > Thanks
> > > >
> > > > Eno
> > > >
> > > >
> > > >
> > > > On Thu, Aug 9, 2018 at 4:18 PM, xiongqi wu 
> > wrote:
> > > >
> > > > > Hi Kafka,
> > > > >
> > > > > This KIP tries to address GDPR concern to fulfill deletion request
> on
> > > > time
> > > > > through time-based log compaction on a compaction enabled topic:
> > > > >
> > > > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> 
> > > > > 354%3A+Time-based+log+compaction+policy
> > > > >
> > > > > Any feedback will be appreciated.
> > > > >
> > > > >
> > > > > Xiongqi (Wesley) Wu
> > > > >
> > > >
> > >
> >
>


-- 

Brett Rann

Senior DevOps Engineer


Zendesk International Ltd

395 Collins Street, Melbourne VIC 3000 Australia

Mobile: +61 (0) 418 826 017

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Brett Rann]

We've been looking into this too.

Mailing list:
https://lists.apache.org/thread.html/ed7f6a6589f94e8c2a705553f364ef599cb6915e4c3ba9b561e610e4@%3Cdev.kafka.apache.org%3E
jira wish: https://issues.apache.org/jira/browse/KAFKA-7137
confluent slack discussion:
https://confluentcommunity.slack.com/archives/C49R61XMM/p153076012139

A person on my team has started on code so you might want to coordinate:
https://github.com/dongxiaohe/kafka/tree/dongxiaohe/log-cleaner-compaction-max-lifetime-2.0

 He's been working with Jason Gustafson and James Chen around the changes.
You can ping him on confluent slack as Xiaohe Dong.

It's great to know others are thinking on it as well.

You've added the requirement to force a segment roll which we hadn't gotten
to yet, which is great. I was content with it not including the active
segment.

> Adding topic level configuration "max.compaction.lag.ms",  and
corresponding broker configuration "log.cleaner.max.compaction.lag.ms",
which is set to 0 (disabled) by default.

Glancing at some other settings convention seems to me to be -1 for
disabled (or infinite, which is more meaningful here).  0 to me implies
instant, a little quicker than 1.

We've been trying to think about a way to trigger compaction as well
through an API call, which would need to be flagged somewhere (ZK admin/
space?) but we're struggling to think how that would be coordinated across
brokers and partitions.  Have you given any thought to that?






On Thu, Aug 16, 2018 at 8:44 AM xiongqi wu  wrote:

> Eno, Dong,
>
> I have updated the KIP. We decide not to address the issue that we might
> have for both compaction and time retention enabled topics (see the
> rejected alternative item 2). This KIP will only ensure log can be
> compacted after a specified time-interval.
>
> As suggested by Dong, we will also enforce "max.compaction.lag.ms" is not
> less than "min.compaction.lag.ms".
>
> https://cwiki.apache.org/confluence/display/KAFKA/KIP-354: Time-based log
> compaction policy
>  compaction policy>
>
>
> On Tue, Aug 14, 2018 at 5:01 PM, xiongqi wu  wrote:
>
> >
> > Per discussion with Dong, he made a very good point that if compaction
> > and time based retention are both enabled on a topic, the compaction
> might
> > prevent records from being deleted on time. The reason is when compacting
> > multiple segments into one single segment, the newly created segment will
> > have same lastmodified timestamp as latest original segment. We lose the
> > timestamp of all original segments except the last one. As a result,
> > records might not be deleted as it should be through time based
> retention.
> >
> > With the current KIP proposal, if we want to ensure timely deletion, we
> > have the following configurations:
> > 1) enable time based log compaction only : deletion is done though
> > overriding the same key
> > 2) enable time based log retention only: deletion is done though
> > time-based retention
> > 3) enable both log compaction and time based retention: Deletion is not
> > guaranteed.
> >
> > Not sure if we have use case 3 and also want deletion to happen on time.
> > There are several options to address deletion issue when enable both
> > compaction and retention:
> > A) During log compaction, looking into record timestamp to delete expired
> > records. This can be done in compaction logic itself or use
> > AdminClient.deleteRecords() . But this assumes we have record timestamp.
> > B) retain the lastModifed time of original segments during log
> compaction.
> > This requires extra meta data to record the information or not grouping
> > multiple segments into one during compaction.
> >
> > If we have use case 3 in general, I would prefer solution A and rely on
> > record timestamp.
> >
> >
> > Two questions:
> > Do we have use case 3? Is it nice to have or must have?
> > If we have use case 3 and want to go with solution A, should we introduce
> > a new configuration to enforce deletion by timestamp?
> >
> >
> > On Tue, Aug 14, 2018 at 1:52 PM, xiongqi wu  wrote:
> >
> >> Dong,
> >>
> >> Thanks for the comment.
> >>
> >> There are two retention policy: log compaction and time based retention.
> >>
> >> Log compaction:
> >>
> >> we have use cases to keep infinite retention of a topic (only
> >> compaction). GDPR cares about deletion of PII (personal identifiable
> >> information) data.
> >> Since Kafka doesn't know what records contain PII, it relies on upper
> >> layer to delete those records.
> >> For those infinite retention uses uses, kafka needs to provide a way to
> >> enforce compaction on time. This is what we try to address in this KIP.
> >>
> >> Time based retention,
> >>
> >> There are also use cases that users of Kafka might want to expire all
> >> their data.
> >> In those cases, they can use time based retention of their topics.
> >>
> >>
> >> Regarding your first question, if a user wants to delete a key 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

Eno, Dong,

I have updated the KIP.  We decide not to address the issue that we might
have for both compaction and time retention enabled topics (see the
rejected alternative item 2).  This KIP will only ensure log can be
compacted after a specified time-interval.

As suggested by Dong,  we will also enforce "max.compaction.lag.ms" is not
less than "min.compaction.lag.ms".

https://cwiki.apache.org/confluence/display/KAFKA/KIP-354%3A+Time-based+log+compaction+policy


On Tue, Aug 14, 2018 at 5:01 PM, xiongqi wu  wrote:

>
> Per discussion with Dong,  he made a very good point that if compaction
> and time based retention are both enabled on a topic, the compaction might
> prevent records from being deleted on time.  The reason is when compacting
> multiple segments into one single segment, the newly created segment will
> have same lastmodified timestamp as latest original segment. We lose the
> timestamp of all original segments except the last one. As a result,
> records might not be deleted as it should be through time based retention.
>
> With the current KIP proposal,   if we want to ensure timely deletion,  we
> have the following configurations:
> 1) enable time based log compaction only :  deletion is done though
> overriding the same key
> 2) enable time based log retention only: deletion is done though
> time-based retention
> 3) enable both log compaction and time based retention:Deletion is not
> guaranteed.
>
> Not sure if we have use case 3 and also want deletion to happen on time.
> There are several options to address deletion issue when enable both
> compaction and retention:
> A) During log compaction, looking into record timestamp to delete expired
> records. This can be done in compaction logic itself or use
> AdminClient.deleteRecords() . But this assumes we have record timestamp.
> B) retain the lastModifed time of original segments during log compaction.
> This requires extra meta data to record the information or not grouping
> multiple segments into one during compaction.
>
> If we have use case 3 in general,  I would prefer solution A and rely on
> record timestamp.
>
>
> Two questions:
> Do we have use case 3? Is it nice to have or must have?
> If we have use case 3 and want to go with solution A,  should we introduce
> a new configuration to enforce deletion by timestamp?
>
>
> On Tue, Aug 14, 2018 at 1:52 PM, xiongqi wu  wrote:
>
>> Dong,
>>
>> Thanks for the comment.
>>
>> There are two retention policy: log compaction and time based retention.
>>
>> Log compaction:
>>
>> we have use cases to keep infinite retention of a topic (only
>> compaction).  GDPR cares about deletion of PII  (personal identifiable
>> information) data.
>> Since Kafka doesn't know what records contain PII, it relies on upper
>> layer to delete those records.
>> For those infinite retention uses uses,  kafka needs to provide a way to
>> enforce compaction on time. This is what we try to address in this KIP.
>>
>> Time based retention,
>>
>> There are also use cases that users of Kafka might want to expire all
>> their data.
>> In those cases, they can use time based retention of their topics.
>>
>>
>> Regarding your first question,  if a user wants to delete a key in the
>> log compaction topic,  the user has to send a deletion using the same key.
>> Kafka only makes sure the deletion will happen under a certain time
>> periods (like 2 days/7 days).
>>
>> Regarding your second question.  In most cases, we might want to delete
>> all duplicated keys at the same time.
>> Compaction might be more efficient since we need to scan the log and find
>> all duplicates.  However,  the expected use case is to set the time based
>> compaction interval on the order of days,  and be larger than 'min
>> compaction lag".  We don't want log compaction to happen frequently since
>> it is expensive.  The purpose is to help low production rate topic to get
>> compacted on time.  For the topic with "normal" incoming message message
>> rate, the "min dirty ratio" might have triggered the compaction before this
>> time based compaction policy takes effect.
>>
>>
>> Eno,
>>
>> For your question,  like I mentioned we have long time retention use case
>> for log compacted topic, but we want to provide ability to delete certain
>> PII records on time.
>> Kafka itself doesn't know whether a record contains sensitive information
>> and relies on the user for deletion.
>>
>>
>> On Mon, Aug 13, 2018 at 6:58 PM, Dong Lin  wrote:
>>
>>> Hey Xiongqi,
>>>
>>> Thanks for the KIP. I have two questions regarding the use-case for
>>> meeting
>>> GDPR requirement.
>>>
>>> 1) If I recall correctly, one of the GDPR requirement is that we can not
>>> keep messages longer than e.g. 30 days in storage (e.g. Kafka). Say there
>>> exists a partition p0 which contains message1 with key1 and message2 with
>>> key2. And then user keeps producing messages with key=key2 to this
>>> partition. Since message1 with key1 is never overridden, sooner or 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

Per discussion with Dong,  he made a very good point that if compaction and
time based retention are both enabled on a topic, the compaction might
prevent records from being deleted on time.  The reason is when compacting
multiple segments into one single segment, the newly created segment will
have same lastmodified timestamp as latest original segment. We lose the
timestamp of all original segments except the last one. As a result,
records might not be deleted as it should be through time based retention.

With the current KIP proposal,   if we want to ensure timely deletion,  we
have the following configurations:
1) enable time based log compaction only :  deletion is done though
overriding the same key
2) enable time based log retention only: deletion is done though time-based
retention
3) enable both log compaction and time based retention:Deletion is not
guaranteed.

Not sure if we have use case 3 and also want deletion to happen on time.
There are several options to address deletion issue when enable both
compaction and retention:
A) During log compaction, looking into record timestamp to delete expired
records. This can be done in compaction logic itself or use
AdminClient.deleteRecords() . But this assumes we have record timestamp.
B) retain the lastModifed time of original segments during log compaction.
This requires extra meta data to record the information or not grouping
multiple segments into one during compaction.

If we have use case 3 in general,  I would prefer solution A and rely on
record timestamp.


Two questions:
Do we have use case 3? Is it nice to have or must have?
If we have use case 3 and want to go with solution A,  should we introduce
a new configuration to enforce deletion by timestamp?


On Tue, Aug 14, 2018 at 1:52 PM, xiongqi wu  wrote:

> Dong,
>
> Thanks for the comment.
>
> There are two retention policy: log compaction and time based retention.
>
> Log compaction:
>
> we have use cases to keep infinite retention of a topic (only
> compaction).  GDPR cares about deletion of PII  (personal identifiable
> information) data.
> Since Kafka doesn't know what records contain PII, it relies on upper
> layer to delete those records.
> For those infinite retention uses uses,  kafka needs to provide a way to
> enforce compaction on time. This is what we try to address in this KIP.
>
> Time based retention,
>
> There are also use cases that users of Kafka might want to expire all
> their data.
> In those cases, they can use time based retention of their topics.
>
>
> Regarding your first question,  if a user wants to delete a key in the log
> compaction topic,  the user has to send a deletion using the same key.
> Kafka only makes sure the deletion will happen under a certain time
> periods (like 2 days/7 days).
>
> Regarding your second question.  In most cases, we might want to delete
> all duplicated keys at the same time.
> Compaction might be more efficient since we need to scan the log and find
> all duplicates.  However,  the expected use case is to set the time based
> compaction interval on the order of days,  and be larger than 'min
> compaction lag".  We don't want log compaction to happen frequently since
> it is expensive.  The purpose is to help low production rate topic to get
> compacted on time.  For the topic with "normal" incoming message message
> rate, the "min dirty ratio" might have triggered the compaction before this
> time based compaction policy takes effect.
>
>
> Eno,
>
> For your question,  like I mentioned we have long time retention use case
> for log compacted topic, but we want to provide ability to delete certain
> PII records on time.
> Kafka itself doesn't know whether a record contains sensitive information
> and relies on the user for deletion.
>
>
> On Mon, Aug 13, 2018 at 6:58 PM, Dong Lin  wrote:
>
>> Hey Xiongqi,
>>
>> Thanks for the KIP. I have two questions regarding the use-case for
>> meeting
>> GDPR requirement.
>>
>> 1) If I recall correctly, one of the GDPR requirement is that we can not
>> keep messages longer than e.g. 30 days in storage (e.g. Kafka). Say there
>> exists a partition p0 which contains message1 with key1 and message2 with
>> key2. And then user keeps producing messages with key=key2 to this
>> partition. Since message1 with key1 is never overridden, sooner or later
>> we
>> will want to delete message1 and keep the latest message with key=key2.
>> But
>> currently it looks like log compact logic in Kafka will always put these
>> messages in the same segment. Will this be an issue?
>>
>> 2) The current KIP intends to provide the capability to delete a given
>> message in log compacted topic. Does such use-case also require Kafka to
>> keep the messages produced before the given message? If yes, then we can
>> probably just use AdminClient.deleteRecords() or time-based log retention
>> to meet the use-case requirement. If no, do you know what is the GDPR's
>> requirement on time-to-deletion after user 

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

Dong,

Thanks for the comment.

There are two retention policy: log compaction and time based retention.

Log compaction:

we have use cases to keep infinite retention of a topic (only compaction).
GDPR cares about deletion of PII  (personal identifiable information) data.
Since Kafka doesn't know what records contain PII, it relies on upper layer
to delete those records.
For those infinite retention uses uses,  kafka needs to provide a way to
enforce compaction on time. This is what we try to address in this KIP.

Time based retention,

There are also use cases that users of Kafka might want to expire all their
data.
In those cases, they can use time based retention of their topics.


Regarding your first question,  if a user wants to delete a key in the log
compaction topic,  the user has to send a deletion using the same key.
Kafka only makes sure the deletion will happen under a certain time periods
(like 2 days/7 days).

Regarding your second question.  In most cases, we might want to delete all
duplicated keys at the same time.
Compaction might be more efficient since we need to scan the log and find
all duplicates.  However,  the expected use case is to set the time based
compaction interval on the order of days,  and be larger than 'min
compaction lag".  We don't want log compaction to happen frequently since
it is expensive.  The purpose is to help low production rate topic to get
compacted on time.  For the topic with "normal" incoming message message
rate, the "min dirty ratio" might have triggered the compaction before this
time based compaction policy takes effect.


Eno,

For your question,  like I mentioned we have long time retention use case
for log compacted topic, but we want to provide ability to delete certain
PII records on time.
Kafka itself doesn't know whether a record contains sensitive information
and relies on the user for deletion.


On Mon, Aug 13, 2018 at 6:58 PM, Dong Lin  wrote:

> Hey Xiongqi,
>
> Thanks for the KIP. I have two questions regarding the use-case for meeting
> GDPR requirement.
>
> 1) If I recall correctly, one of the GDPR requirement is that we can not
> keep messages longer than e.g. 30 days in storage (e.g. Kafka). Say there
> exists a partition p0 which contains message1 with key1 and message2 with
> key2. And then user keeps producing messages with key=key2 to this
> partition. Since message1 with key1 is never overridden, sooner or later we
> will want to delete message1 and keep the latest message with key=key2. But
> currently it looks like log compact logic in Kafka will always put these
> messages in the same segment. Will this be an issue?
>
> 2) The current KIP intends to provide the capability to delete a given
> message in log compacted topic. Does such use-case also require Kafka to
> keep the messages produced before the given message? If yes, then we can
> probably just use AdminClient.deleteRecords() or time-based log retention
> to meet the use-case requirement. If no, do you know what is the GDPR's
> requirement on time-to-deletion after user explicitly requests the deletion
> (e.g. 1 hour, 1 day, 7 day)?
>
> Thanks,
> Dong
>
>
> On Mon, Aug 13, 2018 at 3:44 PM, xiongqi wu  wrote:
>
> > Hi Eno,
> >
> > The GDPR request we are getting here at linkedin is if we get a request
> to
> > delete a record through a null key on a log compacted topic,
> > we want to delete the record via compaction in a given time period like 2
> > days (whatever is required by the policy).
> >
> > There might be other issues (such as orphan log segments under certain
> > conditions)  that lead to GDPR problem but they are more like something
> we
> > need to fix anyway regardless of GDPR.
> >
> >
> > -- Xiongqi (Wesley) Wu
> >
> > On Mon, Aug 13, 2018 at 2:56 PM, Eno Thereska 
> > wrote:
> >
> > > Hello,
> > >
> > > Thanks for the KIP. I'd like to see a more precise definition of what
> > part
> > > of GDPR you are targeting as well as some sort of verification that
> this
> > > KIP actually addresses the problem. Right now I find this a bit vague:
> > >
> > > "Ability to delete a log message through compaction in a timely manner
> > has
> > > become an important requirement in some use cases (e.g., GDPR)"
> > >
> > >
> > > Is there any guarantee that after this KIP the GDPR problem is solved
> or
> > do
> > > we need to do something else as well, e.g., more KIPs?
> > >
> > >
> > > Thanks
> > >
> > > Eno
> > >
> > >
> > >
> > > On Thu, Aug 9, 2018 at 4:18 PM, xiongqi wu 
> wrote:
> > >
> > > > Hi Kafka,
> > > >
> > > > This KIP tries to address GDPR concern to fulfill deletion request on
> > > time
> > > > through time-based log compaction on a compaction enabled topic:
> > > >
> > > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> > > > 354%3A+Time-based+log+compaction+policy
> > > >
> > > > Any feedback will be appreciated.
> > > >
> > > >
> > > > Xiongqi (Wesley) Wu
> > > >
> > >
> >
>



-- 
Xiongqi (Wesley) Wu

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Eno Thereska]

Adding to this, what about topics that are not log compacted? As Dong says,
"one of the GDPR requirement is that we can not keep messages longer than
e.g. 30 days in storage (e.g. Kafka)". The GDPR requirement must hold
irrespective of the low level details, on whether the topic is compacted or
not, right?

Thanks
Eno


On Mon, Aug 13, 2018 at 6:58 PM, Dong Lin  wrote:

> Hey Xiongqi,
>
> Thanks for the KIP. I have two questions regarding the use-case for meeting
> GDPR requirement.
>
> 1) If I recall correctly, one of the GDPR requirement is that we can not
> keep messages longer than e.g. 30 days in storage (e.g. Kafka). Say there
> exists a partition p0 which contains message1 with key1 and message2 with
> key2. And then user keeps producing messages with key=key2 to this
> partition. Since message1 with key1 is never overridden, sooner or later we
> will want to delete message1 and keep the latest message with key=key2. But
> currently it looks like log compact logic in Kafka will always put these
> messages in the same segment. Will this be an issue?
>
> 2) The current KIP intends to provide the capability to delete a given
> message in log compacted topic. Does such use-case also require Kafka to
> keep the messages produced before the given message? If yes, then we can
> probably just use AdminClient.deleteRecords() or time-based log retention
> to meet the use-case requirement. If no, do you know what is the GDPR's
> requirement on time-to-deletion after user explicitly requests the deletion
> (e.g. 1 hour, 1 day, 7 day)?
>
> Thanks,
> Dong
>
>
> On Mon, Aug 13, 2018 at 3:44 PM, xiongqi wu  wrote:
>
> > Hi Eno,
> >
> > The GDPR request we are getting here at linkedin is if we get a request
> to
> > delete a record through a null key on a log compacted topic,
> > we want to delete the record via compaction in a given time period like 2
> > days (whatever is required by the policy).
> >
> > There might be other issues (such as orphan log segments under certain
> > conditions)  that lead to GDPR problem but they are more like something
> we
> > need to fix anyway regardless of GDPR.
> >
> >
> > -- Xiongqi (Wesley) Wu
> >
> > On Mon, Aug 13, 2018 at 2:56 PM, Eno Thereska 
> > wrote:
> >
> > > Hello,
> > >
> > > Thanks for the KIP. I'd like to see a more precise definition of what
> > part
> > > of GDPR you are targeting as well as some sort of verification that
> this
> > > KIP actually addresses the problem. Right now I find this a bit vague:
> > >
> > > "Ability to delete a log message through compaction in a timely manner
> > has
> > > become an important requirement in some use cases (e.g., GDPR)"
> > >
> > >
> > > Is there any guarantee that after this KIP the GDPR problem is solved
> or
> > do
> > > we need to do something else as well, e.g., more KIPs?
> > >
> > >
> > > Thanks
> > >
> > > Eno
> > >
> > >
> > >
> > > On Thu, Aug 9, 2018 at 4:18 PM, xiongqi wu 
> wrote:
> > >
> > > > Hi Kafka,
> > > >
> > > > This KIP tries to address GDPR concern to fulfill deletion request on
> > > time
> > > > through time-based log compaction on a compaction enabled topic:
> > > >
> > > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> > > > 354%3A+Time-based+log+compaction+policy
> > > >
> > > > Any feedback will be appreciated.
> > > >
> > > >
> > > > Xiongqi (Wesley) Wu
> > > >
> > >
> >
>

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Dong Lin]

Hey Xiongqi,

Thanks for the KIP. I have two questions regarding the use-case for meeting
GDPR requirement.

1) If I recall correctly, one of the GDPR requirement is that we can not
keep messages longer than e.g. 30 days in storage (e.g. Kafka). Say there
exists a partition p0 which contains message1 with key1 and message2 with
key2. And then user keeps producing messages with key=key2 to this
partition. Since message1 with key1 is never overridden, sooner or later we
will want to delete message1 and keep the latest message with key=key2. But
currently it looks like log compact logic in Kafka will always put these
messages in the same segment. Will this be an issue?

2) The current KIP intends to provide the capability to delete a given
message in log compacted topic. Does such use-case also require Kafka to
keep the messages produced before the given message? If yes, then we can
probably just use AdminClient.deleteRecords() or time-based log retention
to meet the use-case requirement. If no, do you know what is the GDPR's
requirement on time-to-deletion after user explicitly requests the deletion
(e.g. 1 hour, 1 day, 7 day)?

Thanks,
Dong


On Mon, Aug 13, 2018 at 3:44 PM, xiongqi wu  wrote:

> Hi Eno,
>
> The GDPR request we are getting here at linkedin is if we get a request to
> delete a record through a null key on a log compacted topic,
> we want to delete the record via compaction in a given time period like 2
> days (whatever is required by the policy).
>
> There might be other issues (such as orphan log segments under certain
> conditions)  that lead to GDPR problem but they are more like something we
> need to fix anyway regardless of GDPR.
>
>
> -- Xiongqi (Wesley) Wu
>
> On Mon, Aug 13, 2018 at 2:56 PM, Eno Thereska 
> wrote:
>
> > Hello,
> >
> > Thanks for the KIP. I'd like to see a more precise definition of what
> part
> > of GDPR you are targeting as well as some sort of verification that this
> > KIP actually addresses the problem. Right now I find this a bit vague:
> >
> > "Ability to delete a log message through compaction in a timely manner
> has
> > become an important requirement in some use cases (e.g., GDPR)"
> >
> >
> > Is there any guarantee that after this KIP the GDPR problem is solved or
> do
> > we need to do something else as well, e.g., more KIPs?
> >
> >
> > Thanks
> >
> > Eno
> >
> >
> >
> > On Thu, Aug 9, 2018 at 4:18 PM, xiongqi wu  wrote:
> >
> > > Hi Kafka,
> > >
> > > This KIP tries to address GDPR concern to fulfill deletion request on
> > time
> > > through time-based log compaction on a compaction enabled topic:
> > >
> > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> > > 354%3A+Time-based+log+compaction+policy
> > >
> > > Any feedback will be appreciated.
> > >
> > >
> > > Xiongqi (Wesley) Wu
> > >
> >
>

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

Hi Eno,

The GDPR request we are getting here at linkedin is if we get a request to
delete a record through a null key on a log compacted topic,
we want to delete the record via compaction in a given time period like 2
days (whatever is required by the policy).

There might be other issues (such as orphan log segments under certain
conditions)  that lead to GDPR problem but they are more like something we
need to fix anyway regardless of GDPR.


-- Xiongqi (Wesley) Wu

On Mon, Aug 13, 2018 at 2:56 PM, Eno Thereska 
wrote:

> Hello,
>
> Thanks for the KIP. I'd like to see a more precise definition of what part
> of GDPR you are targeting as well as some sort of verification that this
> KIP actually addresses the problem. Right now I find this a bit vague:
>
> "Ability to delete a log message through compaction in a timely manner has
> become an important requirement in some use cases (e.g., GDPR)"
>
>
> Is there any guarantee that after this KIP the GDPR problem is solved or do
> we need to do something else as well, e.g., more KIPs?
>
>
> Thanks
>
> Eno
>
>
>
> On Thu, Aug 9, 2018 at 4:18 PM, xiongqi wu  wrote:
>
> > Hi Kafka,
> >
> > This KIP tries to address GDPR concern to fulfill deletion request on
> time
> > through time-based log compaction on a compaction enabled topic:
> >
> > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> > 354%3A+Time-based+log+compaction+policy
> >
> > Any feedback will be appreciated.
> >
> >
> > Xiongqi (Wesley) Wu
> >
>

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

Yes.  we want to enforce a max time interval from a message arrival time to
the time the corresponding log segment needs to be compacted.

Today, if the message arriving rate is low for a log compacted topic, the
dirty ratio increases very slowly. As a result, a log segment might be
un-compacted for a long time.

Xiongqi (Welsey) Wu

On Mon, Aug 13, 2018 at 2:46 PM, Guozhang Wang  wrote:

> Guess I need to carefully read the wiki page before asking :) Thanks!
>
> Another qq after reading the proposal: is it a complimentary to KIP-58 (
> https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> 58+-+Make+Log+Compaction+Point+Configurable),
> just that KIP-58 is a "upper-bound" on what messages can be compacted, and
> this is for a "lower-bound" on what messages NEED to be compacted?
>
>
> Guozhang
>
> On Mon, Aug 13, 2018 at 2:31 PM, xiongqi wu  wrote:
>
> > HI Guozhang,
> >
> > As I mentioned in the motivation section, KIP-280 focuses on how to
> compact
> > the log segment to resolve the out of order messages compaction issue.
> > The issue we try to address in this KIP is different:  we want to
> introduce
> > a compaction policy so that a log segment can be pickup for compaction
> > after a specified time interval.  One use case is for GDPR to ensure
> timely
> > deletion of user record.
> >
> > There is no conflict and overlapping between this KIP and KIP-280.
> >
> > Thank you!
> >
> >
> > On Mon, Aug 13, 2018 at 1:33 PM, Guozhang Wang 
> wrote:
> >
> > > Hello Xiongqi,
> > >
> > > I think this KIP is already been covered in KIP-280? Could you check
> out
> > > that one and see if it is the case.
> > >
> > >
> > > Guozhang
> > >
> > >
> > > On Mon, Aug 13, 2018 at 1:23 PM, xiongqi wu 
> wrote:
> > >
> > > > Hi Kafka,
> > > >
> > > > Just updated the confluence page to include the link to this KIP.
> > > >
> > > > Any comment will be appreciated:
> > > >
> > > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-354%3A
> > > > +Time-based+log+compaction+policy
> > > >
> > > > Thank you.
> > > >
> > > > Xiongqi (Wesley) Wu
> > > >
> > > > On Thu, Aug 9, 2018 at 4:18 PM, xiongqi wu 
> > wrote:
> > > >
> > > > > Hi Kafka,
> > > > >
> > > > > This KIP tries to address GDPR concern to fulfill deletion request
> on
> > > > time
> > > > > through time-based log compaction on a compaction enabled topic:
> > > > >
> > > > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> > > > > 354%3A+Time-based+log+compaction+policy
> > > > >
> > > > > Any feedback will be appreciated.
> > > > >
> > > > >
> > > > > Xiongqi (Wesley) Wu
> > > > >
> > > >
> > >
> > >
> > >
> > > --
> > > -- Guozhang
> > >
> >
>
>
>
> --
> -- Guozhang
>

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Eno Thereska]

Hello,

Thanks for the KIP. I'd like to see a more precise definition of what part
of GDPR you are targeting as well as some sort of verification that this
KIP actually addresses the problem. Right now I find this a bit vague:

"Ability to delete a log message through compaction in a timely manner has
become an important requirement in some use cases (e.g., GDPR)"


Is there any guarantee that after this KIP the GDPR problem is solved or do
we need to do something else as well, e.g., more KIPs?


Thanks

Eno



On Thu, Aug 9, 2018 at 4:18 PM, xiongqi wu  wrote:

> Hi Kafka,
>
> This KIP tries to address GDPR concern to fulfill deletion request on time
> through time-based log compaction on a compaction enabled topic:
>
> https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> 354%3A+Time-based+log+compaction+policy
>
> Any feedback will be appreciated.
>
>
> Xiongqi (Wesley) Wu
>

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Guozhang Wang]

Guess I need to carefully read the wiki page before asking :) Thanks!

Another qq after reading the proposal: is it a complimentary to KIP-58 (
https://cwiki.apache.org/confluence/display/KAFKA/KIP-58+-+Make+Log+Compaction+Point+Configurable),
just that KIP-58 is a "upper-bound" on what messages can be compacted, and
this is for a "lower-bound" on what messages NEED to be compacted?


Guozhang

On Mon, Aug 13, 2018 at 2:31 PM, xiongqi wu  wrote:

> HI Guozhang,
>
> As I mentioned in the motivation section, KIP-280 focuses on how to compact
> the log segment to resolve the out of order messages compaction issue.
> The issue we try to address in this KIP is different:  we want to introduce
> a compaction policy so that a log segment can be pickup for compaction
> after a specified time interval.  One use case is for GDPR to ensure timely
> deletion of user record.
>
> There is no conflict and overlapping between this KIP and KIP-280.
>
> Thank you!
>
>
> On Mon, Aug 13, 2018 at 1:33 PM, Guozhang Wang  wrote:
>
> > Hello Xiongqi,
> >
> > I think this KIP is already been covered in KIP-280? Could you check out
> > that one and see if it is the case.
> >
> >
> > Guozhang
> >
> >
> > On Mon, Aug 13, 2018 at 1:23 PM, xiongqi wu  wrote:
> >
> > > Hi Kafka,
> > >
> > > Just updated the confluence page to include the link to this KIP.
> > >
> > > Any comment will be appreciated:
> > >
> > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-354%3A
> > > +Time-based+log+compaction+policy
> > >
> > > Thank you.
> > >
> > > Xiongqi (Wesley) Wu
> > >
> > > On Thu, Aug 9, 2018 at 4:18 PM, xiongqi wu 
> wrote:
> > >
> > > > Hi Kafka,
> > > >
> > > > This KIP tries to address GDPR concern to fulfill deletion request on
> > > time
> > > > through time-based log compaction on a compaction enabled topic:
> > > >
> > > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> > > > 354%3A+Time-based+log+compaction+policy
> > > >
> > > > Any feedback will be appreciated.
> > > >
> > > >
> > > > Xiongqi (Wesley) Wu
> > > >
> > >
> >
> >
> >
> > --
> > -- Guozhang
> >
>



-- 
-- Guozhang

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

HI Guozhang,

As I mentioned in the motivation section, KIP-280 focuses on how to compact
the log segment to resolve the out of order messages compaction issue.
The issue we try to address in this KIP is different:  we want to introduce
a compaction policy so that a log segment can be pickup for compaction
after a specified time interval.  One use case is for GDPR to ensure timely
deletion of user record.

There is no conflict and overlapping between this KIP and KIP-280.

Thank you!


On Mon, Aug 13, 2018 at 1:33 PM, Guozhang Wang  wrote:

> Hello Xiongqi,
>
> I think this KIP is already been covered in KIP-280? Could you check out
> that one and see if it is the case.
>
>
> Guozhang
>
>
> On Mon, Aug 13, 2018 at 1:23 PM, xiongqi wu  wrote:
>
> > Hi Kafka,
> >
> > Just updated the confluence page to include the link to this KIP.
> >
> > Any comment will be appreciated:
> >
> > https://cwiki.apache.org/confluence/display/KAFKA/KIP-354%3A
> > +Time-based+log+compaction+policy
> >
> > Thank you.
> >
> > Xiongqi (Wesley) Wu
> >
> > On Thu, Aug 9, 2018 at 4:18 PM, xiongqi wu  wrote:
> >
> > > Hi Kafka,
> > >
> > > This KIP tries to address GDPR concern to fulfill deletion request on
> > time
> > > through time-based log compaction on a compaction enabled topic:
> > >
> > > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> > > 354%3A+Time-based+log+compaction+policy
> > >
> > > Any feedback will be appreciated.
> > >
> > >
> > > Xiongqi (Wesley) Wu
> > >
> >
>
>
>
> --
> -- Guozhang
>

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[Guozhang Wang]

Hello Xiongqi,

I think this KIP is already been covered in KIP-280? Could you check out
that one and see if it is the case.


Guozhang


On Mon, Aug 13, 2018 at 1:23 PM, xiongqi wu  wrote:

> Hi Kafka,
>
> Just updated the confluence page to include the link to this KIP.
>
> Any comment will be appreciated:
>
> https://cwiki.apache.org/confluence/display/KAFKA/KIP-354%3A
> +Time-based+log+compaction+policy
>
> Thank you.
>
> Xiongqi (Wesley) Wu
>
> On Thu, Aug 9, 2018 at 4:18 PM, xiongqi wu  wrote:
>
> > Hi Kafka,
> >
> > This KIP tries to address GDPR concern to fulfill deletion request on
> time
> > through time-based log compaction on a compaction enabled topic:
> >
> > https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> > 354%3A+Time-based+log+compaction+policy
> >
> > Any feedback will be appreciated.
> >
> >
> > Xiongqi (Wesley) Wu
> >
>



-- 
-- Guozhang

Re: [DISCUSS] KIP-354 Time-based log compaction policy

[xiongqi wu]

Hi Kafka,

Just updated the confluence page to include the link to this KIP.

Any comment will be appreciated:

https://cwiki.apache.org/confluence/display/KAFKA/KIP-354%3A+Time-based+log+compaction+policy

Thank you.

Xiongqi (Wesley) Wu

On Thu, Aug 9, 2018 at 4:18 PM, xiongqi wu  wrote:

> Hi Kafka,
>
> This KIP tries to address GDPR concern to fulfill deletion request on time
> through time-based log compaction on a compaction enabled topic:
>
> https://cwiki.apache.org/confluence/display/KAFKA/KIP-
> 354%3A+Time-based+log+compaction+policy
>
> Any feedback will be appreciated.
>
>
> Xiongqi (Wesley) Wu
>