@Onur - I think you reviewed the KAFKA-2464 patch in some detail. wrt the
specific problem you raised about using largest # of partitions, you're
right, with the updated proposal this isn't possible because we only send a
hash of the metadata in order to avoid the overhead of per-member metadata
being transmitted back to members in the join group response.
However, the KAKFA-2464 patch addresses this by avoiding inconsistencies as
much as possible by preferring the coordinator for metadata fetches when it
detects inconsistencies. If metadata is changing fast enough, this can
still result in another round of inconsistency. However, note that this
only applies to metadata that would be included in an subscription (i.e.,
if quickly created/deleted topics were excluded from mirror maker
consumers, those consumers would not be affected by those metadata changes
since the hash is computed based on the topics included in the
subscription). Additionally, it seems based on Becket's numbers that the
frequency that you *might* hit this is very small and the duration of the
delay in getting the group to reconverge should be 2 rounds of heartbeats,
or 6 seconds based on current defaults.
@Becket - these numbers are *really* useful, thanks for including them. The
average topic name length is actually quite useful it sorting out the
protocol overhead.
1. I think the only reason this is a question is because of the metadata?
Otherwise we always have the standard heartbeat cycle, which is currently
set at 3 seconds. If metadata mismatches, all clients need to refresh.
Because of concerns of inconsistency, the patch for KAFKA-2464 has all
clients in the group hit the coordinator to achieve consistency (only after
they detect inconsistency, so normal case should be spread across brokers).
I'm actually not convinced this is a good idea since it centralizes all
metadata lookups for the group to one host, which seems risky. But it does
guarantee consistency as long as the metadata doesn't change in that
window, and any member that can't fetch metadata should just drop out of
the group anyway. Your numbers seem to suggest it's unlikely to be *that*
unstable in that window. It seems the bad case is that it takes 9 seconds
rather than 6 seconds to converge. And those numbers confirm previous
arguments that the metadata changes slowly enough that a temporary
inconsistency can be resolved very quickly.
2. Rolling bounces are pretty clean regardless of whether assignment is
server- or client-side. We're following the same basic process either way.
In the normal case, processes restart cleanly and upon start, they
immediately try to join the group. There are two cases. If you're the
*first* node to restart, you'll trigger a join group and everyone will need
to rejoin at their next heartbeat. It'll take as long as the heartbeat
interval. If you're not the first node, then there's probably another node
that was bounced that's already triggered the next generation. In that case
you're just waiting to join the next generation anyway, and you'll probably
be done in the next 3 seconds (again, based on current intervals). I think
the major question here is how you schedule the rolling bounce -- if you do
too quickly, you might have a long period where lots of nodes aren't doing
any real work because there's always join groups in progress, which
requires everyone stopping, committing, then starting again. However, this
is true with any version of the current protocol -- it has nothing to do
with client side partition assignment.
Note that this is clearly affected by the proposal for a LeaveGroup
message. Rolling bounce is a case where you might not want to trigger a
LeaveGroup because you actually expect to restart and rejoin the group
immediately. This is a good argument for not using TCP disconnect as a
LeaveGroup signal.
3. I think this is addressed in the previous two comments. Maybe one thing
to address again is that metadata hashes can be based only on topics
matching subscriptions, so any changes in, e.g., test topics that do not
affect a consumer groups subscriptions should have 0 effect. They never
trigger any join groups, never affect membership, etc. Re: actual sync, I'd
be curious if defaulting to the coordinator to ensure consistency has any
problems that you can think of? It's not strictly a guarantee of
consistency since the metadata can change between requests from different
clients, but practically speaking, I don't see a case where metadata would
change so quickly and so consistently that we'd run into problems.
-Ewen
On Fri, Aug 28, 2015 at 9:32 PM, Jiangjie Qin <j...@linkedin.com.invalid>
wrote:
> Hi Neha,
>
> Following are some numbers we have in the pipeline. It would be very
> helpful to see how it goes with the proposed protocol. We will try to do
> some tests with the current patch as well. Please also let us know if you
> want further information.
>
> 32 brokers, 1Gbps NIC
> 547 topics
> 27 chars average topic name length
> 2-3 consumers for each topic
>
> Four 26-node mirror maker instances (Four different consumer groups). Each
> node has 4 consumers. (Each mirror maker instance has 104 consumers)
> We are actually using selective copy, so we have a big whitelist for each
> mirror maker, copying about 100 topics (We expect it to grow to a couple of
> hundreds).
> The mirror makers are co-located with target cluster, so the consumer
> traffic go through the WAN.
>
> We have 5 to 6 wildcard consumers consuming from all the topics.
>
> The topic creation frequency is not high now, roughly about 1 / day.
>
> The scenarios we are interested in are:
> 1. The time for one round of rebalance.
> 2. The time for a rolling bounce of mirror maker.
> 3. For wildcard topic, does metadata sync up cause problem.
>
> Thanks,
>
> Jiangjie (Becket) Qin
>
>
> On Fri, Aug 28, 2015 at 1:24 PM, Joel Koshy <jjkosh...@gmail.com> wrote:
>
> > Another use-case I was thinking of was something like rack-aware
> > assignment of partitions to clients. This would require some
> > additional topic metadata to be propagated to and from the coordinator
> > and you would need some way to resolve conflicts for such strategies.
> > I think that could be addressed by attaching a generation id to the
> > metadata and use that (i.e., pick the highest) in order to resolve
> > conflicts without another round of join-group requests.
> >
> > Likewise, without delete/recreate, partition counts are a sort of
> > generation id since they are non-decreasing. If we need to account for
> > delete/recreate that could perhaps be addressed by an explicit
> > (per-topic) generation id attached to each topic in the metadata blob.
> > Does that make sense? I think that covers my concerns wrt the
> > split-brain issues.
> >
> > I'm still a bit wary of the n^2*m sized rebroadcast of all the
> > metadata - mainly because for various reasons at LinkedIn, we are
> > actually using large explicit whitelists (and not wildcards) in
> > several of our mirroring pipelines. At this point I feel that is a
> > reasonable cost to pay for having all the logic in one place (i.e.,
> > client side) but I would like to think a bit more on that.
> >
> > Joel
> >
> >
> > On Fri, Aug 28, 2015 at 1:02 PM, Onur Karaman
> > <okara...@linkedin.com.invalid> wrote:
> > > From what I understand, the "largest number of partitions" trick is
> based
> > > on the assumption that topics can only expand their partitions. What
> > > happens when a topic gets deleted and recreated? This breaks that
> > > assumption.
> > >
> > > On Fri, Aug 28, 2015 at 6:33 AM, Neha Narkhede <n...@confluent.io>
> > wrote:
> > >
> > >> Thanks for re-reviewing Joel.
> > >>
> > >>
> > >>
> > >>
> > >>
> > >>
> > >> On Fri, Aug 28, 2015 at 2:51 AM -0700, "Joel Koshy" <
> > jjkosh...@gmail.com>
> > >> wrote:
> > >>
> > >>
> > >>
> > >>
> > >>
> > >>
> > >>
> > >>
> > >>
> > >>
> > >> > I think we think this proposal addresses 100% of the split brain
> > issues
> > >> > ever seen in the ZK-based protocol, but I think you think there are
> > still
> > >> > issues. Can you explain what your thinking of and when you think it
> > would
> > >> > happen? I want to make sure you aren't assuming
> > client-side=>split-brain
> > >> > since I think that is totally not the case.
> > >>
> > >> Yes I had concluded that client-side assignment would still result in
> > >> split-brain wrt partition counts, but I overlooked a key sentence in
> > >> the wiki - i.e., that the assignment algorithm for consumers can just
> > >> use the largest number of partitions for each topic reported by any of
> > >> the consumers. i.e., I assumed that consumers would just fail
> > >> rebalance if the partition counts were inconsistent but that is not
> > >> the case since this conflict can be easily resolved as described
> > >> without further join-group requests. Sorry about that. There is still
> > >> the issue of the coordinator having to send back n*m worth of
> > >> metadata, but that was not my biggest concern. I'll look over it again
> > >> and reply back tomorrow.
> > >>
> > >> Joel
> > >>
> > >> On Thu, Aug 27, 2015 at 2:55 PM, Jay Kreps wrote:
> > >> > Hey Joel,
> > >> >
> > >> > I really don't think we should do both. There are pros and cons but
> we
> > >> > should make a decision and work on operationalizing one approach.
> > Much of
> > >> > really making something like this work is getting all the bugs out,
> > >> getting
> > >> > monitoring in place, getting rigorous system tests in place. Trying
> > to do
> > >> > those things twice with the same resources will just mean we do them
> > half
> > >> > as well. I also think this buys nothing from the user's point of
> > >> view--they
> > >> > want co-ordination that works correctly, the debate we are having is
> > >> purely
> > >> > a "how should we build that" debate. So this is really not the kind
> of
> > >> > thing we'd want to make pluggable and if we did that would just
> > >> complicate
> > >> > life for the user.
> > >> >
> > >> > I think we think this proposal addresses 100% of the split brain
> > issues
> > >> > ever seen in the ZK-based protocol, but I think you think there are
> > still
> > >> > issues. Can you explain what your thinking of and when you think it
> > would
> > >> > happen? I want to make sure you aren't assuming
> > client-side=>split-brain
> > >> > since I think that is totally not the case.
> > >> >
> > >> > With respect to "herd issues" I actually think all the proposals
> > address
> > >> > this by scaling the co-ordinator out to all nodes and making the
> > >> > co-ordination vastly cheaper. No proposal, of course, gets rid of
> the
> > >> fact
> > >> > that all clients rejoin at once when there is a membership change,
> but
> > >> that
> > >> > is kind of fundamental to the problem.
> > >> >
> > >> > -Jay
> > >> >
> > >> > On Thu, Aug 27, 2015 at 2:02 PM, Joel Koshy wrote:
> > >> >
> > >> >> I actually feel these set of tests (whatever they may be) are
> > somewhat
> > >> >> irrelevant here. My main concern with the current client-side
> > proposal
> > >> >> (i.e., without Becket's follow-up suggestions) is that it makes a
> > >> >> significant compromise to the original charter of the new consumer
> -
> > >> >> i.e., reduce/eliminate herd and split brain problems in both group
> > >> >> management and partition assignment. I understand the need for
> > >> >> client-side partition assignment in some use cases (which we are
> also
> > >> >> interested in), but I also think we should make every effort to
> keep
> > >> >> full server-side coordination for the remaining (majority) of use
> > >> >> cases especially if it does not complicate the protocol. The
> proposed
> > >> >> changes do not complicate the protocol IMO - i.e., there is no
> > further
> > >> >> modification to the request/response formats beyond the current
> > >> >> client-side proposal. It only involves a trivial reinterpretation
> of
> > >> >> the content of the protocol metadata field.
> > >> >>
> > >> >> Joel
> > >> >>
> > >> >> On Wed, Aug 26, 2015 at 9:33 PM, Neha Narkhede wrote:
> > >> >> > Hey Becket,
> > >> >> >
> > >> >> > In that case, the broker side partition assignment would be ideal
> > >> because
> > >> >> >> it avoids
> > >> >> >> issues like metadata inconsistency / split brain / exploding
> > >> >> subscription
> > >> >> >> set propagation.
> > >> >> >
> > >> >> >
> > >> >> > As per our previous discussions regarding each of those concerns
> > >> >> (referring
> > >> >> > to this email thread, KIP calls and JIRA comments), we are going
> to
> > >> run a
> > >> >> > set of tests using the LinkedIn deployment numbers that we will
> > wait
> > >> for
> > >> >> > you to share. The purpose is to see if those concerns are really
> > >> valid or
> > >> >> > not. I'd prefer to see that before making any more changes that
> > will
> > >> >> > complicate the protocol.
> > >> >> >
> > >> >> > On Wed, Aug 26, 2015 at 4:57 PM, Jiangjie Qin > >
> > >> >> > wrote:
> > >> >> >
> > >> >> >> Hi folks,
> > >> >> >>
> > >> >> >> After further discussion in LinkedIn, we found that while
> having a
> > >> more
> > >> >> >> general group management protocol is very useful, the vast
> > majority
> > >> of
> > >> >> the
> > >> >> >> clients will not use customized partition assignment strategy.
> In
> > >> that
> > >> >> >> case, the broker side partition assignment would be ideal
> because
> > it
> > >> >> avoids
> > >> >> >> issues like metadata inconsistency / split brain / exploding
> > >> >> subscription
> > >> >> >> set propagation.
> > >> >> >>
> > >> >> >> So we have the following proposal that satisfies the majority of
> > the
> > >> >> >> clients' needs without changing the currently proposed binary
> > >> protocol.
> > >> >> >> i.e., Continue to support broker-side assignment if the
> assignment
> > >> >> strategy
> > >> >> >> is recognized by the coordinator.
> > >> >> >>
> > >> >> >> 1. Keep the binary protocol as currently proposed.
> > >> >> >>
> > >> >> >> 2. Change the way we interpret ProtocolMetadata:
> > >> >> >> 2.1 On consumer side, change partition.assignment.strategy to
> > >> >> >> partition.assignor.class. Implement the something like the
> > following
> > >> >> >> PartitionAssignor Interface:
> > >> >> >>
> > >> >> >> public interface PartitionAssignor {
> > >> >> >> List protocolTypes();
> > >> >> >> byte[] protocolMetadata();
> > >> >> >> // return the Topic->List map that are assigned to this
> > >> >> >> consumer.
> > >> >> >> List assignPartitions(String protocolType, byte[]
> > >> >> >> responseProtocolMetadata);
> > >> >> >> }
> > >> >> >>
> > >> >> >> public abstract class AbstractPartitionAssignor implements
> > >> >> >> PartitionAssignor {
> > >> >> >> protected final KafkaConsumer consumer;
> > >> >> >> AbstractPartitionAssignor(KafkaConsumer consumer) {
> > >> >> >> this.consumer = consumer;
> > >> >> >> }
> > >> >> >> }
> > >> >> >>
> > >> >> >> 2.2 The ProtocolMetadata in JoinGroupRequest will be
> > >> >> >> partitionAssignor.protocolMetadata(). When
> > partition.assignor.class
> > >> is
> > >> >> >> "range" or "roundrobin", the ProtocolMetadata in
> JoinGroupRequest
> > >> will
> > >> >> be a
> > >> >> >> JSON subscription set. ("range", "roundrobin" will be reserved
> > >> words, we
> > >> >> >> can also consider reserving some Prefix such as "broker-" to be
> > more
> > >> >> clear)
> > >> >> >> 2.3 On broker side when ProtocolType is "range" or
> "roundroubin",
> > >> >> >> coordinator will parse the ProtocolMetadata in the
> > JoinGroupRequest
> > >> and
> > >> >> >> assign the partitions for consumers. In the JoinGroupResponse,
> the
> > >> >> >> ProtocolMetadata will be the global assignment of partitions.
> > >> >> >> 2.4 On client side, after receiving the JoinGroupResponse,
> > >> >> >> partitionAssignor.assignPartitions() will be invoked to return
> the
> > >> >> actual
> > >> >> >> assignment. If the assignor is RangeAssignor or
> > RoundRobinAssignor,
> > >> they
> > >> >> >> will parse the assignment from the ProtocolMetadata returned by
> > >> >> >> coordinator.
> > >> >> >>
> > >> >> >> This approach has a few merits:
> > >> >> >> 1. Does not change the proposed binary protocol, which is still
> > >> general.
> > >> >> >> 2. The majority of the consumers will not suffer from
> inconsistent
> > >> >> metadata
> > >> >> >> / split brain / exploding subscription set propagation. This is
> > >> >> >> specifically to deal with the issue that the current proposal
> > caters
> > >> to
> > >> >> a
> > >> >> >> 20% use-case while adversely impacting the more common 80%
> > use-cases.
> > >> >> >> 3. Easy to implement. The only thing needed is implement a
> > >> partitioner
> > >> >> >> class. For most users, the default range and roundrobin
> > partitioner
> > >> are
> > >> >> >> good enough.
> > >> >> >>
> > >> >> >> Thoughts?
> > >> >> >>
> > >> >> >> Thanks,
> > >> >> >>
> > >> >> >> Jiangjie (Becket) Qin
> > >> >> >>
> > >> >> >> On Tue, Aug 18, 2015 at 2:51 PM, Jason Gustafson
> > >> >> >> wrote:
> > >> >> >>
> > >> >> >> > Follow-up from the kip call:
> > >> >> >> >
> > >> >> >> > 1. Onur brought up the question of whether this protocol
> > provides
> > >> >> enough
> > >> >> >> > coordination capabilities to be generally useful in practice
> (is
> > >> that
> > >> >> >> > accurate, Onur?). If it doesn't, then each use case would
> > probably
> > >> >> need a
> > >> >> >> > dependence on zookeeper anyway, and we haven't really gained
> > >> anything.
> > >> >> >> The
> > >> >> >> > group membership provided by this protocol is a useful
> primitive
> > >> for
> > >> >> >> > coordination, but it's limited in the sense that everything
> > shared
> > >> >> among
> > >> >> >> > the group has to be communicated at the time the group is
> > created.
> > >> If
> > >> >> any
> > >> >> >> > shared data changes, then the only way the group can ensure
> > >> agreement
> > >> >> is
> > >> >> >> to
> > >> >> >> > force a rebalance. This is expensive since all members must
> > stall
> > >> >> while
> > >> >> >> the
> > >> >> >> > rebalancing takes place. As we have also seen, there is a
> > practical
> > >> >> limit
> > >> >> >> > on the amount of metadata that can be sent through this
> protocol
> > >> when
> > >> >> >> > groups get a little larger. This protocol is therefore not
> > >> suitable to
> > >> >> >> > cases which require frequent communication or which require a
> > large
> > >> >> >> amount
> > >> >> >> > of data to be communicated. For the use cases listed on the
> > wiki,
> > >> >> neither
> > >> >> >> > of these appear to be an issue, but there may be other
> > limitations
> > >> >> which
> > >> >> >> > would limit reuse of the protocol. Perhaps it would be
> > sufficient
> > >> to
> > >> >> >> sketch
> > >> >> >> > how these cases might work?
> > >> >> >> >
> > >> >> >> > 2. We talked a little bit about the issue of metadata churn.
> > Becket
> > >> >> >> brought
> > >> >> >> > up the interesting point that not only do we depend on topic
> > >> metadata
> > >> >> >> > changing relatively infrequently, but we also expect timely
> > >> agreement
> > >> >> >> among
> > >> >> >> > the brokers on what that metadata is. To resolve this, we can
> > have
> > >> the
> > >> >> >> > consumers fetch metadata from the coordinator. We still depend
> > on
> > >> >> topic
> > >> >> >> > metadata not changing frequently, but this should resolve any
> > >> >> >> disagreement
> > >> >> >> > among the brokers themselves. In fact, since we expect that
> > >> >> disagreement
> > >> >> >> is
> > >> >> >> > relatively rare, we can have the consumers fetch from the
> > >> coordinator
> > >> >> >> only
> > >> >> >> > when when a disagreement occurs. The nice thing about this
> > >> proposal is
> > >> >> >> that
> > >> >> >> > it doesn't affect the join group semantics, so the coordinator
> > >> would
> > >> >> >> remain
> > >> >> >> > oblivious to the metadata used by the group for agreement.
> > Also, if
> > >> >> >> > metadata churn becomes an issue, it might be possible to have
> > the
> > >> >> >> > coordinator provide a snapshot for the group to ensure that a
> > >> >> generation
> > >> >> >> > would be able to reach agreement (this would probably require
> > >> adding
> > >> >> >> > groupId/generation to the metadata request).
> > >> >> >> >
> > >> >> >> > 3. We talked briefly about support for multiple protocols in
> the
> > >> join
> > >> >> >> group
> > >> >> >> > request in order to allow changing the assignment strategy
> > without
> > >> >> >> > downtime. I think it's a little doubtful that this would get
> > much
> > >> use
> > >> >> in
> > >> >> >> > practice, but I agree it's a nice option to have on the table.
> > An
> > >> >> >> > alternative, for the sake of argument, is to have each member
> > >> provide
> > >> >> >> only
> > >> >> >> > one version of the protocol, and to let the coordinator choose
> > the
> > >> >> >> protocol
> > >> >> >> > with the largest number of supporters. All members which can't
> > >> support
> > >> >> >> the
> > >> >> >> > selected protocol would be kicked out of the group. The
> drawback
> > >> in a
> > >> >> >> > rolling upgrade is that the total capacity of the group would
> be
> > >> >> >> > momentarily halved. It would also be a little tricky to handle
> > the
> > >> >> case
> > >> >> >> of
> > >> >> >> > retrying when a consumer is kicked out of the group. We
> wouldn't
> > >> want
> > >> >> it
> > >> >> >> to
> > >> >> >> > be able to effect a rebalance, for example, if it would just
> be
> > >> kicked
> > >> >> >> out
> > >> >> >> > again. That would probably complicate the group management
> > logic on
> > >> >> the
> > >> >> >> > coordinator.
> > >> >> >> >
> > >> >> >> >
> > >> >> >> > Thanks,
> > >> >> >> > Jason
> > >> >> >> >
> > >> >> >> >
> > >> >> >> > On Tue, Aug 18, 2015 at 11:16 AM, Jiangjie Qin
> > >> >> > >> >
> > >> >> >> > wrote:
> > >> >> >> >
> > >> >> >> > > Jun,
> > >> >> >> > >
> > >> >> >> > > Yes, I agree. If the metadata can be synced quickly there
> > should
> > >> >> not be
> > >> >> >> > an
> > >> >> >> > > issue. It just occurred to me that there is a proposal to
> > allow
> > >> >> >> consuming
> > >> >> >> > > from followers in ISR, that could potentially cause more
> > frequent
> > >> >> >> > metadata
> > >> >> >> > > change for consumers. Would that be an issue?
> > >> >> >> > >
> > >> >> >> > > Thanks,
> > >> >> >> > >
> > >> >> >> > > Jiangjie (Becket) Qin
> > >> >> >> > >
> > >> >> >> > > On Tue, Aug 18, 2015 at 10:22 AM, Jason Gustafson <
> > >> >> ja...@confluent.io>
> > >> >> >> > > wrote:
> > >> >> >> > >
> > >> >> >> > > > Hi Jun,
> > >> >> >> > > >
> > >> >> >> > > > Answers below:
> > >> >> >> > > >
> > >> >> >> > > > 1. When there are multiple common protocols in the
> > >> >> JoinGroupRequest,
> > >> >> >> > > which
> > >> >> >> > > > one would the coordinator pick?
> > >> >> >> > > >
> > >> >> >> > > > I was intending to use the list to indicate preference. If
> > all
> > >> >> group
> > >> >> >> > > > members support protocols ["A", "B"] in that order, then
> we
> > >> will
> > >> >> >> choose
> > >> >> >> > > > "A." If some support ["B", "A"], then we would either
> choose
> > >> >> based on
> > >> >> >> > > > respective counts or just randomly. The main use case of
> > >> >> supporting
> > >> >> >> the
> > >> >> >> > > > list is for rolling upgrades when a change is made to the
> > >> >> assignment
> > >> >> >> > > > strategy. In that case, the new assignment strategy would
> be
> > >> >> listed
> > >> >> >> > first
> > >> >> >> > > > in the upgraded client. I think it's debatable whether
> this
> > >> >> feature
> > >> >> >> > would
> > >> >> >> > > > get much use in practice, so we might consider dropping
> it.
> > >> >> >> > > >
> > >> >> >> > > > 2. If the protocols don't agree, the group construction
> > fails.
> > >> >> What
> > >> >> >> > > exactly
> > >> >> >> > > > does it mean? Do we send an error in every
> JoinGroupResponse
> > >> and
> > >> >> >> remove
> > >> >> >> > > all
> > >> >> >> > > > members in the group in the coordinator?
> > >> >> >> > > >
> > >> >> >> > > > Yes, that is right. It would be handled similarly to
> > >> inconsistent
> > >> >> >> > > > assignment strategies in the current protocol. The
> > coordinator
> > >> >> >> returns
> > >> >> >> > an
> > >> >> >> > > > error in each join group response, and the client
> propagates
> > >> the
> > >> >> >> error
> > >> >> >> > to
> > >> >> >> > > > the user.
> > >> >> >> > > >
> > >> >> >> > > > 3. Consumer embedded protocol: The proposal has two
> > different
> > >> >> formats
> > >> >> >> > of
> > >> >> >> > > > subscription depending on whether wildcards are used or
> not.
> > >> This
> > >> >> >> > seems a
> > >> >> >> > > > bit complicated. Would it be better to always use the
> > metadata
> > >> >> hash?
> > >> >> >> > The
> > >> >> >> > > > clients know the subscribed topics already. This way, the
> > >> client
> > >> >> code
> > >> >> >> > > > behaves the same whether wildcards are used or not.
> > >> >> >> > > >
> > >> >> >> > > > Yeah, I think this is possible (Neha also suggested it). I
> > >> haven't
> > >> >> >> > > updated
> > >> >> >> > > > the wiki yet, but the patch I started working on uses only
> > the
> > >> >> >> metadata
> > >> >> >> > > > hash. In the case that an explicit topic list is provided,
> > the
> > >> >> hash
> > >> >> >> > just
> > >> >> >> > > > covers the metadata for those topics.
> > >> >> >> > > >
> > >> >> >> > > >
> > >> >> >> > > > Thanks,
> > >> >> >> > > > Jason
> > >> >> >> > > >
> > >> >> >> > > >
> > >> >> >> > > >
> > >> >> >> > > > On Tue, Aug 18, 2015 at 10:06 AM, Jun Rao
> > >> >> wrote:
> > >> >> >> > > >
> > >> >> >> > > > > Jason,
> > >> >> >> > > > >
> > >> >> >> > > > > Thanks for the writeup. A few comments below.
> > >> >> >> > > > >
> > >> >> >> > > > > 1. When there are multiple common protocols in the
> > >> >> >> JoinGroupRequest,
> > >> >> >> > > > which
> > >> >> >> > > > > one would the coordinator pick?
> > >> >> >> > > > > 2. If the protocols don't agree, the group construction
> > >> fails.
> > >> >> What
> > >> >> >> > > > exactly
> > >> >> >> > > > > does it mean? Do we send an error in every
> > JoinGroupResponse
> > >> and
> > >> >> >> > remove
> > >> >> >> > > > all
> > >> >> >> > > > > members in the group in the coordinator?
> > >> >> >> > > > > 3. Consumer embedded protocol: The proposal has two
> > different
> > >> >> >> formats
> > >> >> >> > > of
> > >> >> >> > > > > subscription depending on whether wildcards are used or
> > not.
> > >> >> This
> > >> >> >> > > seems a
> > >> >> >> > > > > bit complicated. Would it be better to always use the
> > >> metadata
> > >> >> >> hash?
> > >> >> >> > > The
> > >> >> >> > > > > clients know the subscribed topics already. This way,
> the
> > >> client
> > >> >> >> code
> > >> >> >> > > > > behaves the same whether wildcards are used or not.
> > >> >> >> > > > >
> > >> >> >> > > > > Jiangjie,
> > >> >> >> > > > >
> > >> >> >> > > > > With respect to rebalance churns due to topics being
> > >> >> >> created/deleted.
> > >> >> >> > > > With
> > >> >> >> > > > > the new consumer, the rebalance can probably settle
> within
> > >> 200ms
> > >> >> >> when
> > >> >> >> > > > there
> > >> >> >> > > > > is a topic change. So, as long as we are not changing
> > topic
> > >> more
> > >> >> >> > than 5
> > >> >> >> > > > > times per sec, there shouldn't be constant churns,
> right?
> > >> >> >> > > > >
> > >> >> >> > > > > Thanks,
> > >> >> >> > > > >
> > >> >> >> > > > > Jun
> > >> >> >> > > > >
> > >> >> >> > > > >
> > >> >> >> > > > >
> > >> >> >> > > > > On Tue, Aug 11, 2015 at 1:19 PM, Jason Gustafson <
> > >> >> >> ja...@confluent.io
> > >> >> >> > >
> > >> >> >> > > > > wrote:
> > >> >> >> > > > >
> > >> >> >> > > > > > Hi Kafka Devs,
> > >> >> >> > > > > >
> > >> >> >> > > > > > One of the nagging issues in the current design of the
> > new
> > >> >> >> consumer
> > >> >> >> > > has
> > >> >> >> > > > > > been the need to support a variety of assignment
> > >> strategies.
> > >> >> >> We've
> > >> >> >> > > > > > encountered this in particular in the design of
> copycat
> > and
> > >> >> the
> > >> >> >> > > > > processing
> > >> >> >> > > > > > framework (KIP-28). From what I understand, Samza also
> > has
> > >> a
> > >> >> >> number
> > >> >> >> > > of
> > >> >> >> > > > > use
> > >> >> >> > > > > > cases with custom assignment needs. The new consumer
> > >> protocol
> > >> >> >> > > supports
> > >> >> >> > > > > new
> > >> >> >> > > > > > assignment strategies by hooking them into the broker.
> > For
> > >> >> many
> > >> >> >> > > > > > environments, this is a major pain and in some cases,
> a
> > >> >> >> > non-starter.
> > >> >> >> > > It
> > >> >> >> > > > > > also challenges the validation that the coordinator
> can
> > >> >> provide.
> > >> >> >> > For
> > >> >> >> > > > > > example, some assignment strategies call for
> partitions
> > to
> > >> be
> > >> >> >> > > assigned
> > >> >> >> > > > > > multiple times, which means that the coordinator can
> > only
> > >> >> check
> > >> >> >> > that
> > >> >> >> > > > > > partitions have been assigned at least once.
> > >> >> >> > > > > >
> > >> >> >> > > > > > To solve these issues, we'd like to propose moving
> > >> assignment
> > >> >> to
> > >> >> >> > the
> > >> >> >> > > > > > client. I've written a wiki which outlines some
> protocol
> > >> >> changes
> > >> >> >> to
> > >> >> >> > > > > achieve
> > >> >> >> > > > > > this:
> > >> >> >> > > > > >
> > >> >> >> > > > > >
> > >> >> >> > > > >
> > >> >> >> > > >
> > >> >> >> > >
> > >> >> >> >
> > >> >> >>
> > >> >>
> > >>
> >
> https://cwiki.apache.org/confluence/display/KAFKA/Kafka+Client-side+Assignment+Proposal
> > >> >> >> > > > > > .
> > >> >> >> > > > > > To summarize briefly, instead of the coordinator
> > assigning
> > >> the
> > >> >> >> > > > partitions
> > >> >> >> > > > > > itself, all subscriptions are forwarded to each member
> > of
> > >> the
> > >> >> >> group
> > >> >> >> > > > which
> > >> >> >> > > > > > then decides independently which partitions it should
> > >> consume.
> > >> >> >> The
> > >> >> >> > > > > protocol
> > >> >> >> > > > > > provides a mechanism for the coordinator to validate
> > that
> > >> all
> > >> >> >> > > consumers
> > >> >> >> > > > > use
> > >> >> >> > > > > > the same assignment strategy, but it does not ensure
> > that
> > >> the
> > >> >> >> > > resulting
> > >> >> >> > > > > > assignment is "correct." This provides a powerful
> > >> capability
> > >> >> for
> > >> >> >> > > users
> > >> >> >> > > > to
> > >> >> >> > > > > > control the full data flow on the client side. They
> > control
> > >> >> how
> > >> >> >> > data
> > >> >> >> > > is
> > >> >> >> > > > > > written to partitions through the Partitioner
> interface
> > and
> > >> >> they
> > >> >> >> > > > control
> > >> >> >> > > > > > how data is consumed through the assignment strategy,
> > all
> > >> >> without
> > >> >> >> > > > > touching
> > >> >> >> > > > > > the server.
> > >> >> >> > > > > >
> > >> >> >> > > > > > Of course nothing comes for free. In particular, this
> > >> change
> > >> >> >> > removes
> > >> >> >> > > > the
> > >> >> >> > > > > > ability of the coordinator to validate that commits
> are
> > >> made
> > >> >> by
> > >> >> >> > > > consumers
> > >> >> >> > > > > > who were assigned the respective partition. This might
> > not
> > >> be
> > >> >> too
> > >> >> >> > bad
> > >> >> >> > > > > since
> > >> >> >> > > > > > we retain the ability to validate the generation id,
> > but it
> > >> >> is a
> > >> >> >> > > > > potential
> > >> >> >> > > > > > concern. We have considered alternative protocols
> which
> > >> add a
> > >> >> >> > second
> > >> >> >> > > > > > round-trip to the protocol in order to give the
> > coordinator
> > >> >> the
> > >> >> >> > > ability
> > >> >> >> > > > > to
> > >> >> >> > > > > > confirm the assignment. As mentioned above, the
> > >> coordinator is
> > >> >> >> > > somewhat
> > >> >> >> > > > > > limited in what it can actually validate, but this
> would
> > >> >> return
> > >> >> >> its
> > >> >> >> > > > > ability
> > >> >> >> > > > > > to validate commits. The tradeoff is that it increases
> > the
> > >> >> >> > protocol's
> > >> >> >> > > > > > complexity which means more ways for the protocol to
> > fail
> > >> and
> > >> >> >> > > > > consequently
> > >> >> >> > > > > > more edge cases in the code.
> > >> >> >> > > > > >
> > >> >> >> > > > > > It also misses an opportunity to generalize the group
> > >> >> membership
> > >> >> >> > > > protocol
> > >> >> >> > > > > > for additional use cases. In fact, after you've gone
> to
> > the
> > >> >> >> trouble
> > >> >> >> > > of
> > >> >> >> > > > > > moving assignment to the client, the main thing that
> is
> > >> left
> > >> >> in
> > >> >> >> > this
> > >> >> >> > > > > > protocol is basically a general group management
> > >> capability.
> > >> >> This
> > >> >> >> > is
> > >> >> >> > > > > > exactly what is needed for a few cases that are
> > currently
> > >> >> under
> > >> >> >> > > > > discussion
> > >> >> >> > > > > > (e.g. copycat or single-writer producer). We've taken
> > this
> > >> >> >> further
> > >> >> >> > > step
> > >> >> >> > > > > in
> > >> >> >> > > > > > the proposal and attempted to envision what that
> general
> > >> >> protocol
> > >> >> >> > > might
> > >> >> >> > > > > > look like and how it could be used both by the
> consumer
> > and
> > >> >> for
> > >> >> >> > some
> > >> >> >> > > of
> > >> >> >> > > > > > these other cases.
> > >> >> >> > > > > >
> > >> >> >> > > > > > Anyway, since time is running out on the new consumer,
> > we
> > >> have
> > >> >> >> > > perhaps
> > >> >> >> > > > > one
> > >> >> >> > > > > > last chance to consider a significant change in the
> > >> protocol
> > >> >> like
> > >> >> >> > > this,
> > >> >> >> > > > > so
> > >> >> >> > > > > > have a look at the wiki and share your thoughts. I've
> no
> > >> doubt
> > >> >> >> that
> > >> >> >> > > > some
> > >> >> >> > > > > > ideas seem clearer in my mind than they do on paper,
> so
> > ask
> > >> >> >> > questions
> > >> >> >> > > > if
> > >> >> >> > > > > > there is any confusion.
> > >> >> >> > > > > >
> > >> >> >> > > > > > Thanks!
> > >> >> >> > > > > > Jason
> > >> >> >> > > > > >
> > >> >> >> > > > >
> > >> >> >> > > >
> > >> >> >> > >
> > >> >> >> >
> > >> >> >>
> > >> >> >
> > >> >> >
> > >> >> >
> > >> >> > --
> > >> >> > Thanks,
> > >> >> > Neha
> > >> >>
> > >>
> >
>
--
Thanks,
Ewen
