Hi Viktor, Thanks for the response. Good questions! answers below: > A few questions regarding the rollback algorithm:> 1. At step 2 how do you > elect the leader?
The step 2 code is in https://github.com/apache/kafka/pull/6296 core/src/main/scala/kafka/controller/KafkaController.scala line#622 rollbackReassignedPartitionLeaderIfRequired(topicPartition, reassignedPartitionContext)During "pending" reassignment, e.g. (1,2,3) => (4,2,5) normally, the leader (in this case broker_id 1) will remain as the leader until all replicas (1,2,3,4,5) in ISR, then the leader will be switched to 4. However, in one scenario, if let's say new replica 4 is already caught up in ISR, and somehow original leader 1 is down or bounced. 4 could become the new leader. rollbackReassignedPartitionLeaderIfRequired() will do a leadership election using PreferredReplicaPartitionLeaderElectionStrategy among brokers in OAR (Original Assigned Replicas set in memory). > 1.1. Would it be always the original leader? Not necessarily, if the original preferred leader is down, it can be other brokers in OAR which are in ISR. > 1.2. What if some brokers that are in OAR are down? If some brokers in OAR are down, the topic/partition will have URP (Under Replicated Partition). The client deciding to do reassignment should be clear what the current state of the cluster is, what brokers are down, what are up, what reassignment is trying to accomplish. e.g. reassignment from down brokers to new brokers(?) > 2. I still have doubts that we need to do the reassignment backwards during > rollback. For instance if we decide to cancel the reassignment at step > #8 > where replicas in OAR - RAR are offline and start the rollback, then how do > we make a replica from OAR online again before electing a leader as described > in step #2 of the rollback algorithm?> 3. Does the algorithm defend against > crashes? Is it able to continue after a controller failover? > 4. I think it would be a good addition if you could add few example scenarios > for rollback. yes. shouldTriggerReassignCancelOnControllerStartup() is the integration test to simulate controller failover while cancelling pending reassignments. I will try to add controller failover scenario in a ducktape system test. You do raise a good point here. If the cluster is in a very BAD shape, e.g. None of the OAR brokers are online, but some new broker in RAR is in ISR and is current leader, it make senses not to rollback to keep that topic/partition online. However, if none of brokers in RAR is online, it may make sense to rollback to OAR and remove it from /admin/reassign_partitions, since the cluster state is already so bad, that topic/partition is offline anyway no matter rollback or not. I will add a check before cancel/rollback a topic/partition's pending reassignment by checking whether at least one broker of OAR is in ISR, so that it can be elected as leader, if not, skip that topic/partition reassignment cancellation and raise an exception. I will list a few more scenarios for rollback. What additional scenarios for rollback you and others can think of? Thanks,George On Monday, February 25, 2019, 3:53:33 AM PST, Viktor Somogyi-Vass <viktorsomo...@gmail.com> wrote: Hey George, Thanks for the prompt response, it makes sense. I'll try to keep your code changes on top of my list and help reviewing that. :)Regarding the incremental reassignment: I don't mind either to discuss it as part of this or in a separate conversation but I think a separate one could be better because both discussions can be long and keeping them separated would limit the scope and make them more digestible and focused. If the community decides to discuss it here then I think I'll put KIP-435 on hold or rejected and add my ideas here. If the community decides to discuss it in a different KIP I think it's a good idea to move the planned future work part into KIP-435 and rework that. Maybe we can co-author it as I think both works could be complementary to each other. In any case I'd be absolutely interested in what others think. A few questions regarding the rollback algorithm:1. At step 2 how do you elect the leader? 1.1. Would it be always the original leader? 1.2. What if some brokers that are in OAR are down?2. I still have doubts that we need to do the reassignment backwards during rollback. For instance if we decide to cancel the reassignment at step #8 where replicas in OAR - RAR are offline and start the rollback, then how do we make a replica from OAR online again before electing a leader as described in step #2 of the rollback algorithm?3. Does the algorithm defend against crashes? Is it able to continue after a controller failover?4. I think it would be a good addition if you could add few example scenarios for rollback. Best, Viktor On Fri, Feb 22, 2019 at 7:04 PM George Li <sql_consult...@yahoo.com> wrote: Hi Viktor, Thanks for reading and provide feedbacks on KIP-236. For reassignments, one can generate a json for new assignments and another json with "original" assignments for rollback purpose. In production cluster, from our experience, we need to submit the reassignments in batches with throttle/staggering to minimize the impact to the cluster. Some large topic/partition couple with throttle can take pretty long time for the new replica to be in ISR to complete reassignment in that batch. Currently during this, Kafka does not allow cancelling the pending reassignments cleanly. Even you have the json with the "original" assignments to rollback, it has to wait till current reassignment to complete, then submit it as reassignments to rollback. If the current reassignment is causing impact to production, we would like the reassignments to be cancelled/rollbacked cleanly/safely/quickly. This is the main goal of KIP-236. The original KIP-236 by Tom Bentley also proposed the incremental reassignments, to submit new reassignments while the current reassignments is still going on. This is scaled back to put under "Planned Future Changes" section of KIP-236, so we can expedite this Reassignment Cancellation/Rollback feature out to the community. The main idea incremental reassignment is to allow submit new reassignments in another ZK node /admin/reassign_partitions_queue and merge it with current pending reassignments in /admin/reassign_partitions. In case of same topic/partition in both ZK node, the conflict resolution is to cancel the current reassignment in /admin/reassign_partitions, and move the same topic/partition from /admin/reassign_partitions_queue as new reassignment. If there is enough interest from the community, this "Planned Future Changes" for incremental reassignments can also be delivered in KIP-236, otherwise, another KIP. The current PR: https://github.com/apache/kafka/pull/6296 only focuses/addresses the pending Reassignment Cancellation/Rollback. Hope this answers your questions. Thanks,George On Friday, February 22, 2019, 6:51:14 AM PST, Viktor Somogyi-Vass <viktorsomo...@gmail.com> wrote: Read through the KIP and I have one comment: It seems like it is not looking strictly for cancellation but also implements rolling back to the original. I think it'd be much simpler to generate a reassignment json on cancellation that contains the original assignment and start a new partition reassignment completely. This way the reassignment algorithm (whatever it is) could be reused as a whole. Did you consider this or are there any obstacles that prevents doing this? Regards, Viktor On Fri, Feb 22, 2019 at 2:24 PM Viktor Somogyi-Vass <viktorsomo...@gmail.com> wrote: > I've published the above mentioned KIP here: > https://cwiki.apache.org/confluence/display/KAFKA/KIP-435%3A+Incremental+Partition+Reassignment > Will start a discussion about it soon. > > On Fri, Feb 22, 2019 at 12:45 PM Viktor Somogyi-Vass < > viktorsomo...@gmail.com> wrote: > >> Hi Folks, >> >> I also have a pending active work on the incremental partition >> reassignment stuff here: https://issues.apache.org/jira/browse/KAFKA-6794 >> I think it would be good to cooperate on this to make both work >> compatible with each other. >> >> I'll write up a KIP about this today so it'll be easier to see how to fit >> the two together. Basically in my work I operate on the >> /admin/reassign_partitions node on a fully compatible way, meaning I won't >> change it just calculate each increment based on that and the current state >> of the ISR set for the partition in reassignment. >> I hope we could collaborate on this. >> >> Viktor >> >> On Thu, Feb 21, 2019 at 9:04 PM Harsha <ka...@harsha.io> wrote: >> >>> Thanks George. LGTM. >>> Jun & Tom, Can you please take a look at the updated KIP. >>> Thanks, >>> Harsha >>> >>> On Wed, Feb 20, 2019, at 12:18 PM, George Li wrote: >>> > Hi, >>> > >>> > After discussing with Tom, Harsha and I are picking up KIP-236 < >>> https://cwiki.apache.org/confluence/display/KAFKA/KIP-236%3A+Interruptible+Partition+Reassignment>. >>> The work focused on safely/cleanly cancel / rollback pending reassignments >>> in a timely fashion. Pull Request #6296 < >>> https://github.com/apache/kafka/pull/6296> Still working on more >>> integration/system tests. >>> > >>> > Please review and provide feedbacks/suggestions. >>> > >>> > Thanks, >>> > George >>> > >>> > >>> > On Saturday, December 23, 2017, 0:51:13 GMT, Jun Rao <j...@confluent.io> >>> wrote: >>> > >>> > Hi, Tom, >>> >>> Thanks for the reply. >>> >>> 10. That's a good thought. Perhaps it's better to get rid of >>> /admin/reassignment_requests >>> too. The window when a controller is not available is small. So, we can >>> just failed the admin client if the controller is not reachable after the >>> timeout. >>> >>> 13. With the changes in 10, the old approach is handled through ZK >>> callback >>> and the new approach is through Kafka RPC. The ordering between the two >>> is >>> kind of arbitrary. Perhaps the ordering can just be based on the order >>> that >>> the reassignment is added to the controller request queue. From there, we >>> can either do the overriding or the prevention. >>> >>> Jun >>> >>> >>> On Fri, Dec 22, 2017 at 7:31 AM, Tom Bentley <t.j.bent...@gmail.com> >>> wrote: >>> >>> > Hi Jun, >>> > >>> > Thanks for responding, my replies are inline: >>> > >>> > 10. You explanation makes sense. My remaining concern is the >>> additional ZK >>> > > writes in the proposal. With the proposal, we will need to do >>> following >>> > > writes in ZK. >>> > > >>> > > a. write new assignment in /admin/reassignment_requests >>> > > >>> > > b. write new assignment and additional metadata in >>> > > /admin/reassignments/$topic/$partition >>> > > >>> > > c. write old + new assignment in /brokers/topics/[topic] >>> > > >>> > > d. write new assignment in /brokers/topics/[topic] >>> > > >>> > > e. delete /admin/reassignments/$topic/$partition >>> > > >>> > > So, there are quite a few ZK writes. I am wondering if it's better to >>> > > consolidate the info in /admin/reassignments/$topic/$partition into >>> > > /brokers/topics/[topic]. >>> > > For example, we can just add some new JSON fields in >>> > > /brokers/topics/[topic] >>> > > to remember the new assignment and potentially the original replica >>> count >>> > > when doing step c. Those fields with then be removed in step d. That >>> way, >>> > > we can get rid of step b and e, saving 2 ZK writes per partition. >>> > > >>> > >>> > This seems like a great idea to me. >>> > >>> > It might also be possible to get rid of the >>> /admin/reassignment_requests >>> > subtree too. I've not yet published the ideas I have for the >>> AdminClient >>> > API for reassigning partitions, but given the existence of such an >>> API, the >>> > route to starting a reassignment would be the AdminClient, and not >>> > zookeeper. In that case there is no need for >>> /admin/reassignment_requests >>> > at all. The only drawback that I can see is that while it's currently >>> > possible to trigger a reassignment even during a controller >>> > election/failover that would no longer be the case if all requests had >>> to >>> > go via the controller. >>> > >>> > >>> > > 11. What you described sounds good. We could potentially optimize the >>> > > dropped replicas a bit more. Suppose that assignment [0,1,2] is first >>> > > changed to [1,2,3] and then to [2,3,4]. When initiating the second >>> > > assignment, we may end up dropping replica 3 and only to restart it >>> > again. >>> > > In this case, we could only drop a replica if it's not going to be >>> added >>> > > back again. >>> > > >>> > >>> > I had missed that, thank you! I will update the proposed algorithm to >>> > prevent this. >>> > >>> > >>> > > 13. Since this is a corner case, we can either prevent or allow >>> > overriding >>> > > with old/new mechanisms. To me, it seems that allowing is simpler to >>> > > implement, the order in /admin/reassignment_requests determines the >>> > > ordering the of override, whether that's initiated by the new way or >>> the >>> > > old way. >>> > > >>> > >>> > That makes sense except for the corner case where: >>> > >>> > * There is no current controller and >>> > * Writes to both the new and old znodes happen >>> > >>> > On election of the new controller, for those partitions with both a >>> > reassignment_request and in /admin/reassign_partitions, we have to >>> decide >>> > which should win. You could use the modification time, though there are >>> > some very unlikely scenarios where that doesn't work properly, for >>> example >>> > if both znodes have the same mtime, or the /admin/reassign_partitions >>> was >>> > updated, but the assignment of the partition wasn't changed, like this: >>> > >>> > 0. /admin/reassign_partitions has my-topic/42 = [1,2,3] >>> > 1. Controller stops watching. >>> > 2. Create /admin/reassignment_requests/request_1234 to change the >>> > reassignment of partition my-topic/42 = [4,5,6] >>> > 3. Update /admin/reassign_partitions to add your-topic/12=[7,8,9] >>> > 4. New controller resumes >>> > >>> > >>> > >>> > > Thanks, >>> > > >>> > > Jun >>> > > >>> > > On Tue, Dec 19, 2017 at 2:43 AM, Tom Bentley <t.j.bent...@gmail.com> >>> > > wrote: >>> > > >>> > > > Hi Jun, >>> > > > >>> > > > 10. Another concern of mine is on consistency with the current >>> pattern. >>> > > The >>> > > > > current pattern for change notification based on ZK is (1) we >>> first >>> > > write >>> > > > > the actual value in the entity path and then write the change >>> > > > notification >>> > > > > path, and (2) the change notification path only includes what >>> entity >>> > > has >>> > > > > changed but not the actual changes. If we want to follow this >>> pattern >>> > > for >>> > > > > consistency, /admin/reassignment_requests/request_xxx will only >>> have >>> > > the >>> > > > > partitions whose reassignment have changed, but not the actual >>> > > > > reassignment. >>> > > > > >>> > > > >>> > > > Ah, I hadn't understood part (2). That means my concern about >>> > efficiency >>> > > > with the current pattern is misplaced. There are still some >>> interesting >>> > > > differences in semantics, however: >>> > > > >>> > > > a) The mechanism currently proposed in KIP-236 means that the >>> > controller >>> > > is >>> > > > the only writer to /admin/reassignments. This means it can include >>> > > > information in these znodes that requesters might not know, or >>> > > information >>> > > > that's necessary to perform the reassignment but not necessary to >>> > > describe >>> > > > the request. While this could be handled using the current pattern >>> it >>> > > would >>> > > > rely on all writers to preserve any information added by the >>> > controller, >>> > > > which seems complicated and hence fragile. >>> > > > >>> > > > b) The current pattern for change notification doesn't cope with >>> > > competing >>> > > > writers to the entity path: If two processes write to the entity >>> path >>> > > > before the controller can read it (due to notification) then one >>> set of >>> > > > updates will be lost. >>> > > > >>> > > > c) If a single writing process crashes after writing to the entity >>> > path, >>> > > > but before writing to the notification path then the write will be >>> > lost. >>> > > > >>> > > > I'm actually using point a) in my WIP (see below). Points b) and >>> c) are >>> > > > obviously edge cases. >>> > > > >>> > > > >>> > > > > 11. Ok. I am not sure that I fully understand the description of >>> that >>> > > > part. >>> > > > > Does "assigned" refer to the current assignment? Could you also >>> > > describe >>> > > > > where the length of the original assignment is stored in ZK? >>> > > > > >>> > > > >>> > > > Sorry if the description is not clear. Yes, "assigned" referrs to >>> the >>> > > > currently assigned replicas (taken from the >>> > > > ControllerContext.partitionReplicaAssignment). I would store the >>> > length >>> > > of >>> > > > the original assignment in the >>> /admin/reassignments/$topic/$partition >>> > > > znode >>> > > > (this is where the point (a) above is useful -- the requester >>> shouldn't >>> > > > know that this information is used by the controller). >>> > > > >>> > > > I've updated the KIP to make these points clearer. >>> > > > >>> > > > >>> > > > > 13. Hmm, I am not sure that the cancellation needs to be done >>> for the >>> > > > whole >>> > > > > batch. The reason that I brought this up is for consistency. The >>> KIP >>> > > > allows >>> > > > > override when using the new approach. It just seems that it's >>> simpler >>> > > to >>> > > > > extend this model when resolving multiple changes between the >>> old and >>> > > the >>> > > > > new approach. >>> > > > >>> > > > >>> > > > Ah, I think I've been unclear on this point too. Currently the >>> > > > ReassignPartitionsCommand enforces that you can't change >>> reassignments, >>> > > but >>> > > > this doesn't stop other ZK clients making changes to >>> > > > /admin/reassign_partitions directly and I believe some Kafka users >>> do >>> > > > indeed change reassignments in-flight by writing to >>> > > > /admin/reassign_partitions. What I'm proposing doesn't break that >>> at >>> > all. >>> > > > The semantic I've implemented is only that the controller only >>> refuses >>> > a >>> > > > reassignment change if there is already one in-flight (i.e. in >>> > > > /admin/reassignments/$topic/$partition) **via the other >>> mechansim**. >>> > So >>> > > if >>> > > > you're using /admin/reassign_partitions and you change or cancel >>> part >>> > of >>> > > it >>> > > > via /admin/reassign_partitions, that's OK. Likewise if you're using >>> > > > /admin/reassignment_request/request_xxx and you change or cancel >>> part >>> > of >>> > > > it >>> > > > via another /admin/reassignment_request/request_xxx, that's OK.What >>> > you >>> > > > can't do is change a request that was started via >>> > > > /admin/reassign_partitions via /admin/reassignment_request/ >>> > request_xxx, >>> > > or >>> > > > vice versa. >>> > > > >>> > > > What I was thinking of when I replied is the case where, on >>> controller >>> > > > failover, /admin/reassign_partitions has been changed and >>> > > > /admin/reassignment_request/request_xxx created (in the period when >>> > the >>> > > > new >>> > > > controller was being elected, for example) with a common >>> partition. In >>> > > this >>> > > > case we should apply a consistent rule to that used when the >>> > notification >>> > > > happen in real time. Your suggestion to use the modification time >>> of >>> > the >>> > > > znode would work here too (except in the edge case where ZK writes >>> to >>> > > both >>> > > > znodes happen within the same clock tick on the ZK server, so the >>> > mtimes >>> > > > are the same). >>> > > > >>> > > > Let me know if you think this is the right semantic and I'll try to >>> > > clarify >>> > > > the KIP. >>> > > > >>> > > > Many thanks, >>> > > > >>> > > > Tom >>> > > > >>> > > > On 18 December 2017 at 18:12, Jun Rao <j...@confluent.io> wrote: >>> > > > >>> > > > > Hi, Tom, >>> > > > > >>> > > > > Thanks for the reply. A few more followup comments below. >>> > > > > >>> > > > > 10. Another concern of mine is on consistency with the current >>> > pattern. >>> > > > The >>> > > > > current pattern for change notification based on ZK is (1) we >>> first >>> > > write >>> > > > > the actual value in the entity path and then write the change >>> > > > notification >>> > > > > path, and (2) the change notification path only includes what >>> entity >>> > > has >>> > > > > changed but not the actual changes. If we want to follow this >>> pattern >>> > > for >>> > > > > consistency, /admin/reassignment_requests/request_xxx will only >>> have >>> > > the >>> > > > > partitions whose reassignment have changed, but not the actual >>> > > > > reassignment. >>> > > > > >>> > > > > 11. Ok. I am not sure that I fully understand the description of >>> that >>> > > > part. >>> > > > > Does "assigned" refer to the current assignment? Could you also >>> > > describe >>> > > > > where the length of the original assignment is stored in ZK? >>> > > > > >>> > > > > 13. Hmm, I am not sure that the cancellation needs to be done >>> for the >>> > > > whole >>> > > > > batch. The reason that I brought this up is for consistency. The >>> KIP >>> > > > allows >>> > > > > override when using the new approach. It just seems that it's >>> simpler >>> > > to >>> > > > > extend this model when resolving multiple changes between the >>> old and >>> > > the >>> > > > > new approach. >>> > > > > >>> > > > > Jun >>> > > > > >>> > > > > >>> > > > > >>> > > > > On Mon, Dec 18, 2017 at 2:45 AM, Tom Bentley < >>> t.j.bent...@gmail.com> >>> > > > > wrote: >>> > > > > >>> > > > > > Hi Jun, >>> > > > > > >>> > > > > > Thanks for replying, some answers below: >>> > > > > > >>> > > > > > >>> > > > > > > 10. The proposal now stores the reassignment for all >>> partitions >>> > in >>> > > > > > > /admin/reassignment_requests/request_xxx. If the number of >>> > > > reassigned >>> > > > > > > partitions is larger, the ZK write may hit the default 1MB >>> limit >>> > > and >>> > > > > > fail. >>> > > > > > > An alternative approach is to have the reassignment requester >>> > first >>> > > > > write >>> > > > > > > the new assignment for each partition under >>> > > > > > > /admin/reassignments/$topic/$partition and then write >>> > > > > > > /admin/reassignment_requests/request_xxx with an empty value. >>> > The >>> > > > > > > controller can then read all values under >>> /admin/reassignments. >>> > > > > > > >>> > > > > > >>> > > > > > You're right that reassigning enough partitions would hit the >>> 1MB >>> > > > limit, >>> > > > > > but I don't think this would be a problem in practice because >>> it >>> > > would >>> > > > be >>> > > > > > trivial to split the partitions into several requests (i.e. >>> > mutleiple >>> > > > > > request_xxx). >>> > > > > > I don't think the non-atomicity this would imply is a problem. >>> By >>> > > > writing >>> > > > > > the partitions whose /admin/reassignments/$topic/$partition has >>> > been >>> > > > > > created or changed it makes it much more efficient to know >>> which of >>> > > > those >>> > > > > > znodes we need to read. If I understand your suggestion, you >>> would >>> > > have >>> > > > > to >>> > > > > > read every node under /admin/reassignments to figure out which >>> had >>> > > > > changed. >>> > > > > > >>> > > > > > >>> > > > > > > 11. The improvement you suggested in >>> onPartitionReassignment() >>> > > sounds >>> > > > > > good >>> > > > > > > at the high level. The computation of those dropped >>> partitions >>> > > seems >>> > > > a >>> > > > > > bit >>> > > > > > > complicated. Perhaps a simple approach is to drop the >>> replicas >>> > not >>> > > in >>> > > > > the >>> > > > > > > original assignment and newest reassignment? >>> > > > > > > >>> > > > > > >>> > > > > > This was what I came up with originally too, but when I looked >>> into >>> > > > > > implementing it I found a couple of things which made me >>> reconsider >>> > > it. >>> > > > > > Consider the reassignments [0,1] -> [2,3] -> [3,4]. In words: >>> we >>> > > start >>> > > > > > reassigning to [2,3], but then change our minds about 2 and >>> switch >>> > it >>> > > > to >>> > > > > 4 >>> > > > > > (maybe we've figured out a better overall balance). At that >>> point >>> > it >>> > > is >>> > > > > > perfectly possible that broker 2 is in-sync and broker 1 is not >>> > > > in-sync. >>> > > > > It >>> > > > > > seems silly to drop broker 2 in favour of broker 1: We're >>> > needlessly >>> > > > > giving >>> > > > > > the cluster more work to do. >>> > > > > > >>> > > > > > The second thing that made me reconsider was in that same >>> scenario >>> > > it's >>> > > > > > even possible that broker 2 is the leader of the partition. >>> > Obviously >>> > > > we >>> > > > > > can elect a new leader before dropping it, but not without >>> causing >>> > > > > > disruption to producers and consumers. >>> > > > > > >>> > > > > > By accepting a little more complexity in choosing which >>> brokers to >>> > > drop >>> > > > > we >>> > > > > > make the dropping simpler (no need for leader election) and >>> ensure >>> > > the >>> > > > > > cluster has less work to do. >>> > > > > > >>> > > > > > >>> > > > > > > 12. You brought up the need of remembering the original >>> > assignment. >>> > > > > This >>> > > > > > > will be lost if the assignment is changed multiple times if >>> we >>> > > follow >>> > > > > the >>> > > > > > > approach described in 10. One way is to store the original >>> > > assignment >>> > > > > in >>> > > > > > > /brokers/topics/[topic] as the following. When the final >>> > > reassignment >>> > > > > > > completes, we can remove the original field. >>> > > > > > > { >>> > > > > > > "version": 1, >>> > > > > > > "partitions": {"0": [0, 1, 3] }, >>> > > > > > > "originals": {"0": [0, 1, 2] } >>> > > > > > > } >>> > > > > > > >>> > > > > > >>> > > > > > While I was implementing my first version of >>> > > onPartitionReassignment(), >>> > > > > > where I preferred the originals, I was storing the originals >>> in the >>> > > > > > /admin/reassignments/$topic/$partition znodes. Since we will >>> > remove >>> > > > that >>> > > > > > znode at the end of reassignment anyway, I would suggest this >>> is a >>> > > > better >>> > > > > > place to store that data (if it's necessary to do so), so that >>> we >>> > can >>> > > > > avoid >>> > > > > > another ZK round trip. >>> > > > > > >>> > > > > > >>> > > > > > > 13. For resolving the conflict between >>> /admin/reassign_partitions >>> > > and >>> > > > > > > /admin/reassignments/$topic/$partition, perhaps it's more >>> > natural >>> > > to >>> > > > > > just >>> > > > > > > let the assignment with a newer timestamp to override the >>> older >>> > > one? >>> > > > > > > >>> > > > > > >>> > > > > > That would work but with slightly different semantics to what I >>> > have: >>> > > > > Since >>> > > > > > /admin/reassign_partitions contains multiple partitions, using >>> the >>> > > > > > timestamp means the whole batch wins or losses. By tracking how >>> > each >>> > > > > > request was made we can be more fine-grained. I'm to use the >>> > > > modification >>> > > > > > time if such granularity is not required. >>> > > > > > >>> > > > > > >>> > > > > > > 14. Implementation wise, currently, we register a watcher of >>> the >>> > > isr >>> > > > > path >>> > > > > > > of each partition being reassigned. This has the potential >>> issue >>> > of >>> > > > > > > registering many listeners. An improvement could be just >>> > > piggybacking >>> > > > > on >>> > > > > > > the existing IsrChangeNotificationHandler, which only >>> watches a >>> > > > single >>> > > > > ZK >>> > > > > > > path and is triggered on a batch of isr changes. This is >>> kind of >>> > > > > > orthogonal >>> > > > > > > to the KIP. However, if we are touching the reassignment >>> logic, >>> > it >>> > > > may >>> > > > > be >>> > > > > > > worth considering. >>> > > > > > >>> > > > > > >>> > > > > > Let me look into that. >>> > > > > > >>> > > > > > Thanks, >>> > > > > > >>> > > > > > Tom >>> > > > > > >>> > > > > > On 16 December 2017 at 02:19, Jun Rao <j...@confluent.io> >>> wrote: >>> > > > > > >>> > > > > > > Hi, Tom, >>> > > > > > > >>> > > > > > > Thanks for the updated KIP. A few more comments below. >>> > > > > > > >>> > > > > > > 10. The proposal now stores the reassignment for all >>> partitions >>> > in >>> > > > > > > /admin/reassignment_requests/request_xxx. If the number of >>> > > > reassigned >>> > > > > > > partitions is larger, the ZK write may hit the default 1MB >>> limit >>> > > and >>> > > > > > fail. >>> > > > > > > An alternative approach is to have the reassignment requester >>> > first >>> > > > > write >>> > > > > > > the new assignment for each partition under >>> > > > > > > /admin/reassignments/$topic/$partition and then write >>> > > > > > > /admin/reassignment_requests/request_xxx with an empty value. >>> > The >>> > > > > > > controller can then read all values under >>> /admin/reassignments. >>> > > > > > > >>> > > > > > > 11. The improvement you suggested in >>> onPartitionReassignment() >>> > > sounds >>> > > > > > good >>> > > > > > > at the high level. The computation of those dropped >>> partitions >>> > > seems >>> > > > a >>> > > > > > bit >>> > > > > > > complicated. Perhaps a simple approach is to drop the >>> replicas >>> > not >>> > > in >>> > > > > the >>> > > > > > > original assignment and newest reassignment? >>> > > > > > > >>> > > > > > > 12. You brought up the need of remembering the original >>> > assignment. >>> > > > > This >>> > > > > > > will be lost if the assignment is changed multiple times if >>> we >>> > > follow >>> > > > > the >>> > > > > > > approach described in 10. One way is to store the original >>> > > assignment >>> > > > > in >>> > > > > > > /brokers/topics/[topic] as the following. When the final >>> > > reassignment >>> > > > > > > completes, we can remove the original field. >>> > > > > > > { >>> > > > > > > "version": 1, >>> > > > > > > "partitions": {"0": [0, 1, 3] }, >>> > > > > > > "originals": {"0": [0, 1, 2] } >>> > > > > > > } >>> > > > > > > >>> > > > > > > 13. For resolving the conflict between >>> /admin/reassign_partitions >>> > > and >>> > > > > > > /admin/reassignments/$topic/$partition, perhaps it's more >>> > natural >>> > > to >>> > > > > > just >>> > > > > > > let the assignment with a newer timestamp to override the >>> older >>> > > one? >>> > > > > > > >>> > > > > > > 14. Implementation wise, currently, we register a watcher of >>> the >>> > > isr >>> > > > > path >>> > > > > > > of each partition being reassigned. This has the potential >>> issue >>> > of >>> > > > > > > registering many listeners. An improvement could be just >>> > > piggybacking >>> > > > > on >>> > > > > > > the existing IsrChangeNotificationHandler, which only >>> watches a >>> > > > single >>> > > > > ZK >>> > > > > > > path and is triggered on a batch of isr changes. This is >>> kind of >>> > > > > > orthogonal >>> > > > > > > to the KIP. However, if we are touching the reassignment >>> logic, >>> > it >>> > > > may >>> > > > > be >>> > > > > > > worth considering. >>> > > > > > > >>> > > > > > > Thanks, >>> > > > > > > >>> > > > > > > Jun >>> > > > > > > >>> > > > > > > On Fri, Dec 15, 2017 at 10:17 AM, Tom Bentley < >>> > > t.j.bent...@gmail.com >>> > > > > >>> > > > > > > wrote: >>> > > > > > > >>> > > > > > > > Just wanted to mention that I've started KIP-240, which >>> builds >>> > on >>> > > > top >>> > > > > > of >>> > > > > > > > this one to provide an AdminClient API for listing and >>> > describing >>> > > > > > > > reassignments. >>> > > > > > > > >>> > > > > > > > On 15 December 2017 at 14:34, Tom Bentley < >>> > t.j.bent...@gmail.com >>> > > > >>> > > > > > wrote: >>> > > > > > > > >>> > > > > > > > > > Should we seek to improve this algorithm in this KIP, >>> or >>> > > leave >>> > > > > that >>> > > > > > > as >>> > > > > > > > > a later optimisation? >>> > > > > > > > > >>> > > > > > > > > I've updated the KIP with a proposed algorithm. >>> > > > > > > > > >>> > > > > > > > > >>> > > > > > > > > >>> > > > > > > > > >>> > > > > > > > > >>> > > > > > > > > >>> > > > > > > > > On 14 December 2017 at 09:57, Tom Bentley < >>> > > t.j.bent...@gmail.com >>> > > > > >>> > > > > > > wrote: >>> > > > > > > > > >>> > > > > > > > >> Thanks Ted, now fixed. >>> > > > > > > > >> >>> > > > > > > > >> On 13 December 2017 at 18:38, Ted Yu < >>> yuzhih...@gmail.com> >>> > > > wrote: >>> > > > > > > > >> >>> > > > > > > > >>> Tom: >>> > > > > > > > >>> bq. create a znode >>> /admin/reassignments/$topic-$partition >>> > > > > > > > >>> >>> > > > > > > > >>> Looks like the tree structure above should be: >>> > > > > > > > >>> >>> > > > > > > > >>> /admin/reassignments/$topic/$partition >>> > > > > > > > >>> >>> > > > > > > > >>> bq. The controller removes /admin/reassignment/$topic/$ >>> > > > partition >>> > > > > > > > >>> >>> > > > > > > > >>> Note the lack of 's' for reassignment. It would be >>> good to >>> > > make >>> > > > > > > > zookeeper >>> > > > > > > > >>> paths consistent. >>> > > > > > > > >>> >>> > > > > > > > >>> Thanks >>> > > > > > > > >>> >>> > > > > > > > >>> On Wed, Dec 13, 2017 at 9:49 AM, Tom Bentley < >>> > > > > > t.j.bent...@gmail.com> >>> > > > > > > > >>> wrote: >>> > > > > > > > >>> >>> > > > > > > > >>> > Hi Jun and Ted, >>> > > > > > > > >>> > >>> > > > > > > > >>> > Jun, you're right that needing one watcher per >>> reassigned >>> > > > > > partition >>> > > > > > > > >>> > presents a scalability problem, and using a separate >>> > > > > notification >>> > > > > > > > path >>> > > > > > > > >>> > solves that. I also agree that it makes sense to >>> prevent >>> > > > users >>> > > > > > from >>> > > > > > > > >>> using >>> > > > > > > > >>> > both methods on the same reassignment. >>> > > > > > > > >>> > >>> > > > > > > > >>> > Ted, naming the reassignments like mytopic-42 was >>> simpler >>> > > > > while I >>> > > > > > > was >>> > > > > > > > >>> > proposing a watcher-per-reassignment (I'd have >>> needed a >>> > > child >>> > > > > > > watcher >>> > > > > > > > >>> on >>> > > > > > > > >>> > /admin/reassignments and also on >>> > > > /admin/reassignments/mytopic). >>> > > > > > > Using >>> > > > > > > > >>> the >>> > > > > > > > >>> > separate notification path means I don't need any >>> > watchers >>> > > in >>> > > > > the >>> > > > > > > > >>> > /admin/reassignments subtree, so switching to >>> > > > > > > > >>> /admin/reassignments/mytopic/ >>> > > > > > > > >>> > 42 >>> > > > > > > > >>> > would work, and avoid /admin/reassignments having a >>> very >>> > > > large >>> > > > > > > number >>> > > > > > > > >>> of >>> > > > > > > > >>> > child nodes. On the other hand it also means I have >>> to >>> > > create >>> > > > > and >>> > > > > > > > >>> delete >>> > > > > > > > >>> > the topic nodes (e.g. /admin/reassignments/mytopic), >>> > which >>> > > > > incurs >>> > > > > > > the >>> > > > > > > > >>> cost >>> > > > > > > > >>> > of extra round trips to zookeeper. I suppose that >>> since >>> > > > > > > reassignment >>> > > > > > > > is >>> > > > > > > > >>> > generally a slow process it makes little difference >>> if we >>> > > > > > increase >>> > > > > > > > the >>> > > > > > > > >>> > latency of the interactions with zookeeper. >>> > > > > > > > >>> > >>> > > > > > > > >>> > I have updated the KIP with these improvements, and a >>> > more >>> > > > > > detailed >>> > > > > > > > >>> > description of exactly how we would manage these >>> znodes. >>> > > > > > > > >>> > >>> > > > > > > > >>> > Reading the algorithm in KafkaController. >>> > > > > > > onPartitionReassignment(), >>> > > > > > > > it >>> > > > > > > > >>> > seems that it would be suboptimal for changing >>> > > reassignments >>> > > > > > > > in-flight. >>> > > > > > > > >>> > Consider an initial assignment of [1,2], reassigned >>> to >>> > > [2,3] >>> > > > > and >>> > > > > > > then >>> > > > > > > > >>> > changed to [2,4]. Broker 3 will remain in the >>> assigned >>> > > > replicas >>> > > > > > > until >>> > > > > > > > >>> > broker 4 is in sync, even though 3 wasn't actually >>> one of >>> > > the >>> > > > > > > > original >>> > > > > > > > >>> > assigned replicas and is no longer a new assigned >>> > replica. >>> > > I >>> > > > > > think >>> > > > > > > > this >>> > > > > > > > >>> > also affects the case where the reassignment is >>> cancelled >>> > > > > > > > >>> > ([1,2]->[2,3]->[1,2]): We again have to wait for 3 to >>> > catch >>> > > > up, >>> > > > > > > even >>> > > > > > > > >>> though >>> > > > > > > > >>> > its replica will then be deleted. >>> > > > > > > > >>> > >>> > > > > > > > >>> > Should we seek to improve this algorithm in this >>> KIP, or >>> > > > leave >>> > > > > > that >>> > > > > > > > as >>> > > > > > > > >>> a >>> > > > > > > > >>> > later optimisation? >>> > > > > > > > >>> > >>> > > > > > > > >>> > Cheers, >>> > > > > > > > >>> > >>> > > > > > > > >>> > Tom >>> > > > > > > > >>> > >>> > > > > > > > >>> > On 11 December 2017 at 21:31, Jun Rao < >>> j...@confluent.io> >>> > > > > wrote: >>> > > > > > > > >>> > >>> > > > > > > > >>> > > Another question is on the compatibility. Since now >>> > there >>> > > > > are 2 >>> > > > > > > > ways >>> > > > > > > > >>> of >>> > > > > > > > >>> > > specifying a partition reassignment, one under >>> > > > > > > > >>> /admin/reassign_partitions >>> > > > > > > > >>> > > and the other under /admin/reassignments, we >>> probably >>> > > want >>> > > > to >>> > > > > > > > >>> prevent the >>> > > > > > > > >>> > > same topic being reassigned under both paths at the >>> > same >>> > > > > time? >>> > > > > > > > >>> > > Thanks, >>> > > > > > > > >>> > > >>> > > > > > > > >>> > > Jun >>> > > > > > > > >>> > > >>> > > > > > > > >>> > > >>> > > > > > > > >>> > > >>> > > > > > > > >>> > > On Fri, Dec 8, 2017 at 5:41 PM, Jun Rao < >>> > > j...@confluent.io> >>> > > > > > > wrote: >>> > > > > > > > >>> > > >>> > > > > > > > >>> > > > Hi, Tom, >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > > Thanks for the KIP. It definitely addresses one >>> of >>> > the >>> > > > pain >>> > > > > > > > points >>> > > > > > > > >>> in >>> > > > > > > > >>> > > > partition reassignment. Another issue that it >>> also >>> > > > > addresses >>> > > > > > is >>> > > > > > > > >>> the ZK >>> > > > > > > > >>> > > node >>> > > > > > > > >>> > > > size limit when writing the reassignment JSON. >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > > My only concern is that the KIP needs to create >>> one >>> > > > watcher >>> > > > > > per >>> > > > > > > > >>> > > reassigned >>> > > > > > > > >>> > > > partition. This could add overhead in ZK and >>> > complexity >>> > > > for >>> > > > > > > > >>> debugging >>> > > > > > > > >>> > > when >>> > > > > > > > >>> > > > lots of partitions are being reassigned >>> > simultaneously. >>> > > > We >>> > > > > > > could >>> > > > > > > > >>> > > > potentially improve this by introducing a >>> separate ZK >>> > > > path >>> > > > > > for >>> > > > > > > > >>> change >>> > > > > > > > >>> > > > notification as we do for configs. For example, >>> every >>> > > > time >>> > > > > we >>> > > > > > > > >>> change >>> > > > > > > > >>> > the >>> > > > > > > > >>> > > > assignment for a set of partitions, we could >>> further >>> > > > write >>> > > > > a >>> > > > > > > > >>> sequential >>> > > > > > > > >>> > > > node /admin/reassignment_changes/[change_x]. That >>> > way, >>> > > > the >>> > > > > > > > >>> controller >>> > > > > > > > >>> > > > only needs to watch the change path. Once a >>> change is >>> > > > > > > triggered, >>> > > > > > > > >>> the >>> > > > > > > > >>> > > > controller can read everything under >>> > > > /admin/reassignments/. >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > > Jun >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > > On Wed, Dec 6, 2017 at 1:19 PM, Tom Bentley < >>> > > > > > > > t.j.bent...@gmail.com >>> > > > > > > > >>> > >>> > > > > > > > >>> > > wrote: >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > >> Hi, >>> > > > > > > > >>> > > >> >>> > > > > > > > >>> > > >> This is still very new, but I wanted some quick >>> > > feedback >>> > > > > on >>> > > > > > a >>> > > > > > > > >>> > > preliminary >>> > > > > > > > >>> > > >> KIP which could, I think, help with providing an >>> > > > > AdminClient >>> > > > > > > API >>> > > > > > > > >>> for >>> > > > > > > > >>> > > >> partition reassignment. >>> > > > > > > > >>> > > >> >>> > > > > > > > >>> > > >> https://cwiki.apache.org/ >>> > > confluence/display/KAFKA/KIP- >>> > > > > 236% >>> > > > > > > > >>> > > >> 3A+Interruptible+Partition+Reassignment >>> > > > > > > > >>> > > >> >>> > > > > > > > >>> > > >> I wasn't sure whether to start fleshing out a >>> whole >>> > > > > > > AdminClient >>> > > > > > > > >>> API in >>> > > > > > > > >>> > > >> this >>> > > > > > > > >>> > > >> KIP (which would make it very big, and >>> difficult to >>> > > > read), >>> > > > > > or >>> > > > > > > > >>> whether >>> > > > > > > > >>> > to >>> > > > > > > > >>> > > >> break it down into smaller KIPs (which makes it >>> > easier >>> > > > to >>> > > > > > read >>> > > > > > > > and >>> > > > > > > > >>> > > >> implement in pieces, but harder to get a >>> high-level >>> > > > > picture >>> > > > > > of >>> > > > > > > > the >>> > > > > > > > >>> > > >> ultimate >>> > > > > > > > >>> > > >> destination). For now I've gone for a very small >>> > > initial >>> > > > > > KIP, >>> > > > > > > > but >>> > > > > > > > >>> I'm >>> > > > > > > > >>> > > >> happy >>> > > > > > > > >>> > > >> to sketch the bigger picture here if people are >>> > > > > interested. >>> > > > > > > > >>> > > >> >>> > > > > > > > >>> > > >> Cheers, >>> > > > > > > > >>> > > >> >>> > > > > > > > >>> > > >> Tom >>> > > > > > > > >>> > > >> >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > >>> > > > > > > > >>> > >>> > > > > > > > >>> > >>> > > > > > > > >>> > On 11 December 2017 at 21:31, Jun Rao < >>> j...@confluent.io> >>> > > > > wrote: >>> > > > > > > > >>> > >>> > > > > > > > >>> > > Another question is on the compatibility. Since now >>> > there >>> > > > > are 2 >>> > > > > > > > ways >>> > > > > > > > >>> of >>> > > > > > > > >>> > > specifying a partition reassignment, one under >>> > > > > > > > >>> /admin/reassign_partitions >>> > > > > > > > >>> > > and the other under /admin/reassignments, we >>> probably >>> > > want >>> > > > to >>> > > > > > > > >>> prevent the >>> > > > > > > > >>> > > same topic being reassigned under both paths at the >>> > same >>> > > > > time? >>> > > > > > > > >>> > > Thanks, >>> > > > > > > > >>> > > >>> > > > > > > > >>> > > Jun >>> > > > > > > > >>> > > >>> > > > > > > > >>> > > >>> > > > > > > > >>> > > >>> > > > > > > > >>> > > On Fri, Dec 8, 2017 at 5:41 PM, Jun Rao < >>> > > j...@confluent.io> >>> > > > > > > wrote: >>> > > > > > > > >>> > > >>> > > > > > > > >>> > > > Hi, Tom, >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > > Thanks for the KIP. It definitely addresses one >>> of >>> > the >>> > > > pain >>> > > > > > > > points >>> > > > > > > > >>> in >>> > > > > > > > >>> > > > partition reassignment. Another issue that it >>> also >>> > > > > addresses >>> > > > > > is >>> > > > > > > > >>> the ZK >>> > > > > > > > >>> > > node >>> > > > > > > > >>> > > > size limit when writing the reassignment JSON. >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > > My only concern is that the KIP needs to create >>> one >>> > > > watcher >>> > > > > > per >>> > > > > > > > >>> > > reassigned >>> > > > > > > > >>> > > > partition. This could add overhead in ZK and >>> > complexity >>> > > > for >>> > > > > > > > >>> debugging >>> > > > > > > > >>> > > when >>> > > > > > > > >>> > > > lots of partitions are being reassigned >>> > simultaneously. >>> > > > We >>> > > > > > > could >>> > > > > > > > >>> > > > potentially improve this by introducing a >>> separate ZK >>> > > > path >>> > > > > > for >>> > > > > > > > >>> change >>> > > > > > > > >>> > > > notification as we do for configs. For example, >>> every >>> > > > time >>> > > > > we >>> > > > > > > > >>> change >>> > > > > > > > >>> > the >>> > > > > > > > >>> > > > assignment for a set of partitions, we could >>> further >>> > > > write >>> > > > > a >>> > > > > > > > >>> sequential >>> > > > > > > > >>> > > > node /admin/reassignment_changes/[change_x]. That >>> > way, >>> > > > the >>> > > > > > > > >>> controller >>> > > > > > > > >>> > > > only needs to watch the change path. Once a >>> change is >>> > > > > > > triggered, >>> > > > > > > > >>> the >>> > > > > > > > >>> > > > controller can read everything under >>> > > > /admin/reassignments/. >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > > Jun >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > > On Wed, Dec 6, 2017 at 1:19 PM, Tom Bentley < >>> > > > > > > > t.j.bent...@gmail.com >>> > > > > > > > >>> > >>> > > > > > > > >>> > > wrote: >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > >> Hi, >>> > > > > > > > >>> > > >> >>> > > > > > > > >>> > > >> This is still very new, but I wanted some quick >>> > > feedback >>> > > > > on >>> > > > > > a >>> > > > > > > > >>> > > preliminary >>> > > > > > > > >>> > > >> KIP which could, I think, help with providing an >>> > > > > AdminClient >>> > > > > > > API >>> > > > > > > > >>> for >>> > > > > > > > >>> > > >> partition reassignment. >>> > > > > > > > >>> > > >> >>> > > > > > > > >>> > > >> https://cwiki.apache.org/ >>> > > confluence/display/KAFKA/KIP- >>> > > > > 236% >>> > > > > > > > >>> > > >> 3A+Interruptible+Partition+Reassignment >>> > > > > > > > >>> > > >> >>> > > > > > > > >>> > > >> I wasn't sure whether to start fleshing out a >>> whole >>> > > > > > > AdminClient >>> > > > > > > > >>> API in >>> > > > > > > > >>> > > >> this >>> > > > > > > > >>> > > >> KIP (which would make it very big, and >>> difficult to >>> > > > read), >>> > > > > > or >>> > > > > > > > >>> whether >>> > > > > > > > >>> > to >>> > > > > > > > >>> > > >> break it down into smaller KIPs (which makes it >>> > easier >>> > > > to >>> > > > > > read >>> > > > > > > > and >>> > > > > > > > >>> > > >> implement in pieces, but harder to get a >>> high-level >>> > > > > picture >>> > > > > > of >>> > > > > > > > the >>> > > > > > > > >>> > > >> ultimate >>> > > > > > > > >>> > > >> destination). For now I've gone for a very small >>> > > initial >>> > > > > > KIP, >>> > > > > > > > but >>> > > > > > > > >>> I'm >>> > > > > > > > >>> > > >> happy >>> > > > > > > > >>> > > >> to sketch the bigger picture here if people are >>> > > > > interested. >>> > > > > > > > >>> > > >> >>> > > > > > > > >>> > > >> Cheers, >>> > > > > > > > >>> > > >> >>> > > > > > > > >>> > > >> Tom >>> > > > > > > > >>> > > >> >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > > >>> > > > > > > > >>> > > >>> > > > > > > > >>> > >>> > > > > > > > >>> >>> > > > > > > > >> >>> > > > > > > > >> >>> > > > > > > > > >>> > > > > > > > >>> > > > > > > >>> > > > > > >>> > > > > >>> > > > >>> > > >>> > >>> > >>> >>
