Hi Andrew, about AS30, I had a closer look at KIP-939, and I think this is what you mean:
The problem is that a WriteTxnMarker request for a PID unknown to the ProducerStateManager (cleared by MIRROR_PID_RESET) does not fail. The end transaction marker is still appended, but transaction index and LSO are unaffected (ProducerStateManager.prepareUpdate, ProducerAppendInfo.appendEndTxnMarker). This may lead to a situation where we have two different end txn markers for the same transaction: ABORT from the stop mirroring operation, followed by a COMMIT from the local Txn Coordinator retrying its WriteTxnMarker after the partition becomes writable again. READ_COMMITTED consumers still see the earlier ABORT in the transaction index and skip the aborted data. The visible outcome for consumers is correct. However, __transaction_state on the original source cluster now records the transaction as COMMITTED while the partition log has it as aborted, creating a silent inconsistency that tools querying transaction state will not detect. The window for this race is bounded by transaction.max.timeout.ms (default 15 minutes), but becomes unbounded with KIP-939 because it allows prepared transactions to sit indefinitely. We could move the coordinator state to “complete abort” at the cost of an RPC. This would prevent any retry, but that would break the 2PC contract (Kafka must hold the prepared state until the external coordinator decides). Also, before stop there would be a number of retries constantly going on. We definitely need a better solution. On Thu, Jun 18, 2026 at 11:36 AM Federico Valeri <[email protected]> wrote: > > Hi Andrew, thanks for your careful review. > > AS28: We reworked that section and added some diagrams to illustrate > problem and solutions. Let us know if it is better. > > AS30: We updated that table, but we are not sure about the problem you > mention. Can you clarify and send us the sequence of actions that > would lead to that problem? > > AS32: You are right. Replaced StartMirrorTopicsRequestData.TopicData > with a Controller.MirrorTopicMetadata record. Also renamed TopicData > to TopicMetadata in all schemas. > > AS33: Good idea. Added per-topic error messages in the RPC responses > you mentioned. > > AS29, AS31, AS34 (both of them), AS35, AS36: These small issues are > all fixed now. > > > > > On Tue, Jun 9, 2026 at 9:58 PM Andrew Schofield <[email protected]> wrote: > > > > Hi Fede and friends, > > This KIP is shaping up nicely. The change to the way that state management > > for the mirrors is done is much more elegant. The removal of the PID > > mapping is also a nice improvement. > > > > Inevitably, I have some more comments. > > > > AS28: I understand KAFKA-18723 and how leader epochs work in that case, but > > the enhancements that you made to the Fetch RPC and how they work escapes > > me. Please could we have some diagrams to illustrate the various epochs? > > > > AS29: I like the addition of share group offset syncing. You need > > AlterShareGroupOffsets and DescribeShareGroupOffsets in the tables of RPCs > > and permissions. > > > > AS30: It is certainly true that EOS is not supported. Using cluster > > mirroring on transactional data does risk breaking transactional atomicity > > for transactions in flight at the time that mirroring was stopped. That's a > > known limitation, and it's an effect of per-topic async replication. > > Writing ABORT markers when stop mirroring is triggered is interesting. I > > just wonder whether it's truly safe. > > > > Could you enhance the tables of records in the Exactly-Once Semantics > > section to include leader epochs? I think that the leader epoch for the > > ABORT markers will have been bumped as the mirroring stopped and the > > partitions became writable. However, I worry that someone might switch the > > mirroring back in the other direction before the transaction which was > > rudely partially aborted has completed on the original source cluster, and > > that the effect of the transaction coordinator completing the transaction, > > potentially writing duplicate transaction markers after the ABORT markers > > which will have been replicated back again. > > > > This is probably only a realistic possibility when KIP-939 is present, but > > it is accepted and I'm sure it will land soon enough. I would say this KIP > > should be prepared for the existence of long-running transactions so we > > don't dig a hole to fall into later. > > > > AS31: In the section on the Admin Client, the description for > > Admin.createClusterMirror(String, Map, CreateClusterMirrorOptions) is > > duplicated. > > > > AS32: The use of StartMirrorTopicsRequestData.TopicData in an external > > interface looks odd. > > > > AS33: I suggest per-topic error messages in the RPC responses such as > > PauseMirrorTopics so you can pass back more context, such as the state if > > the request failed due to a state mismatch. > > > > AS34: There is an unnamed field in DescribeClusterMirrorsResponse. The same > > in LastMirrorEpochsValue. > > > > AS34: Some of the fields in BumpLeaderEpochsRequest start with lower-case > > letters. Also, there are quite a few fields in the RPCs in general that > > should be version "0+" which are "0". > > > > AS35: Sometimes the partition states in the RPCs are string, but sometimes > > int8. > > > > AS36: The base level for share group support is probably 4.1. All of the > > RPCs were present in 4.1, although they were only enabled by default in 4.2. > > > > Thanks, > > Andrew > > > > On 2026/06/04 11:01:36 Federico Valeri wrote: > > > Hello, additional updates and replies: > > > > > > VK12: Consumer data loss risk is a good point. We updated the "Group > > > Offsets" paragraph. > > > > > > Some of you raised the point that using mirror.name suffixes for pause > > > and stop operations is a bit inelegant and we agreed. Now we propose a > > > better approach: > > > > > > We replaced the topic config based approach (mirror.name and > > > .stopped/.paused suffix conventions) with a first class metadata > > > record for tracking mirror topic state changes. A new > > > MirrorTopicStateChangeRecord is added to the metadata log with three > > > fields: TopicId (uuid), MirrorName (string), and DesiredState (int8, > > > where 0 = start mirroring, 1 = stop, 2 = pause). When a user calls > > > startMirrorTopics, stopMirrorTopics, or pauseMirrorTopics, the > > > controller writes this record to the metadata log instead of > > > manipulating config suffixes. Brokers receive the record via metadata > > > updates and react accordingly, triggering partition level state > > > transitions through the existing MirrorPartitionState state machine. > > > More details in the updated KIP. > > > > > > Let us know what you think. > > > > > > Thanks > > > Fede > > > > > > > > > On Tue, Jun 2, 2026 at 8:50 AM Luke Chen <[email protected]> wrote: > > > > > > > > Hi Viquar, > > > > > > > > Thanks for the comment. > > > > > > > > Regarding VK12, thanks for raising the issue. > > > > We didn't think about that. > > > > I agree that compared with data re-processing, it's worse to have data > > > > loss. > > > > The proposed formula makes sense to me. > > > > *syncedOffset = max(destinationLogStartOffset, min(sourceCommitted, > > > > destinationLEO))* > > > > > > > > Let us have some discussion internally and then update the KIP and > > > > reply to > > > > the thread. > > > > > > > > Thanks, > > > > Luke > > > > > > > > On Wed, May 20, 2026 at 2:39 AM Federico Valeri <[email protected]> > > > > wrote: > > > > > > > > > Hi Rajini, > > > > > > > > > > RS14: Done. Missed that, sorry. > > > > > > > > > > Thanks again. > > > > > > > > > > On Tue, May 19, 2026 at 7:28 PM Rajini Sivaram > > > > > <[email protected]> > > > > > wrote: > > > > > > > > > > > > Hi Federico, > > > > > > > > > > > > Thanks for the updates. Just one minor point, apart from that, looks > > > > > good. > > > > > > > > > > > > RS14: KIP still shows "bin/kafka-configs.sh --bootstrap-server :9094 > > > > > > --entity-type mirrors". > > > > > > Will be good to change that to `--entity-type cluster-mirrors`. > > > > > > > > > > > > Thanks, > > > > > > > > > > > > Rajini > > > > > > > > > > > > > > > > > > On Tue, May 19, 2026 at 5:28 PM vaquar khan <[email protected]> > > > > > wrote: > > > > > > > > > > > > > Hi Federico and team, > > > > > > > > > > > > > > Thank you for your detailed response on May 11, 2026. I greatly > > > > > appreciate > > > > > > > the collaborative effort over the past few months to harden the > > > > > KIP-1279 > > > > > > > architecture. > > > > > > > > > > > > > > * 1. Resolved Architectural & Stability Vulnerabilities* > > > > > > > > > > > > > > I am pleased to confirm that the KIP has successfully integrated > > > > > > > fixes > > > > > for > > > > > > > the critical vulnerabilities I flagged, thereby protecting the > > > > > cluster's > > > > > > > state machine, memory bounds, and control plane. Consider the > > > > > > > following > > > > > > > items fully resolved on my end: > > > > > > > > > > > > > > - > > > > > > > > > > > > > > *VK4 / VK8 (Negative PID Bug & State Machine Failure):* I > > > > > > > previously > > > > > > > identified that the proposed -(sourceProducerId + 2) PID > > > > > > > rewriting > > > > > > > formula would fundamentally break hasProducerId() in > > > > > > > AbstractRecordBatch.java, causing the broker to bypass > > > > > > > ProducerStateManager.update() and default the Last Stable > > > > > > > Offset > > > > > (LSO) > > > > > > > to the High Watermark. I am glad to see the KIP authors > > > > > > > abandoned > > > > > PID > > > > > > > mapping entirely and adopted the deterministic MIRROR_PID_RESET > > > > > control > > > > > > > record barrier. This perfectly protects Kafka's exactly-once > > > > > semantics. > > > > > > > > > > > > > > - > > > > > > > > > > > > > > *VK11 (TransactionIndex Rebuild Ambiguity):* Thank you for > > > > > officially > > > > > > > confirming that the TransactionIndex is strictly rebuilt > > > > > > > locally > > > > > during > > > > > > > log appends rather than copied byte-for-byte. This resolves my > > > > > concern > > > > > > > regarding PID mismatches inducing consumer read anomalies. > > > > > > > - > > > > > > > > > > > > > > *VK1 (Thundering Herd / OOM Heap Allocation):* Your > > > > > > > clarification > > > > > that > > > > > > > fetcher threads multiplex partitions meaning the memory > > > > > > > footprint is > > > > > > > strictly bounded by num_fetcher_threads * response_max_bytes > > > > > > > rather > > > > > than > > > > > > > a concurrent 1MB buffer per partition fully resolves my concern > > > > > > > regarding > > > > > > > 50GB broker-wide OOM spikes during mass partition wake-ups. > > > > > > > - > > > > > > > > > > > > > > *VK3 (Control Plane Hotspots):* I had flagged the severe risk > > > > > > > of > > > > > > > metadata saturation on a single broker during a "link flap" > > > > > > > event. > > > > > Your > > > > > > > confirmation that the __mirror_state topic utilizes a compound > > > > > > > hash > > > > > > > of (mirrorName, > > > > > > > topicId, partition number) mathematically ensures that the > > > > > > > 50,000+ > > > > > state > > > > > > > transitions will be safely distributed across the cluster, > > > > > neutralizing > > > > > > > the > > > > > > > single-node hotspot risk. > > > > > > > > > > > > > > 2. Outstanding Critical Blocker: VK12 (Offset Sync Data Loss) > > > > > > > > > > > > > > Regarding VK12, there is a fundamental misunderstanding in your > > > > > previous > > > > > > > reply. You stated: *"The scenario described can only occur if > > > > > > > offsets > > > > > are > > > > > > > force-written to an active group, which the design prevents."* > > > > > > > > > > > > > > My concern has absolutely nothing to do with overwriting active > > > > > groups. My > > > > > > > concern applies strictly to the normal synchronization of > > > > > > > inactive/dead > > > > > > > groups, and is based directly on the race condition currently > > > > > documented in > > > > > > > the official KIP-1279 text: > > > > > > > > > > > > > > *"During offset synchronization, the committed offset in the > > > > > destination > > > > > > > cluster may temporarily exceed the current log end offset (LEO) > > > > > > > of the > > > > > > > mirror topic... consumers attempting to resume from offset 100 > > > > > > > will > > > > > receive > > > > > > > an OffsetOutOfRangeException. To handle this gracefully, consumers > > > > > should > > > > > > > configure auto.offset.reset=latest..."* > > > > > > > > > > > > > > If a failover happens during this documented divergence window, a > > > > > > > reconnecting consumer will hit the OffsetOutOfRangeException. If > > > > > > > the > > > > > > > downstream consumer follows the KIP's official advice and relies > > > > > > > on > > > > > > > auto.offset.reset=latest, the consumer will jump to the absolute > > > > > > > newest > > > > > > > offset on the partition. *This completely skips any newly produced > > > > > records > > > > > > > that arrived between the failover and the consumer reconnecting, > > > > > resulting > > > > > > > in silent, unrecoverable data loss for the downstream > > > > > > > application.* > > > > > > > *Proposed Resolution: Double-Clamped Offset Safety Invariant* > > > > > > > Instead > > > > > of > > > > > > > requiring consumers to use auto.offset.reset=latest and endorsing > > > > > > > a > > > > > known > > > > > > > data loss vector, I propose that the ClusterMirrorCoordinator > > > > > > > enforce a > > > > > > > pre-persist offset validation invariant during offset > > > > > > > synchronization. > > > > > > > Before any translated offset is committed to the destination > > > > > > > cluster, > > > > > the > > > > > > > coordinator must apply a double-clamp: > > > > > > > > > > > > > > *syncedOffset = max(destinationLSO, min(sourceCommitted, > > > > > destinationLEO))* > > > > > > > > > > > > > > Where destinationLSO is the Log Start Offset (earliest readable > > > > > position > > > > > > > post-retention) and destinationLEO is the Log End Offset (latest > > > > > replicated > > > > > > > position) of the destination partition. > > > > > > > > > > > > > > This guarantees that every persisted offset falls within the > > > > > > > physically > > > > > > > valid range eliminating both the OffsetOutOfRangeException > > > > > > > caused by > > > > > > > exceeding the LEO during replication lag, and the expired-offset > > > > > > > rewind > > > > > > > caused by falling below the LSO due to destination retention > > > > > > > policies. > > > > > > > > > > > > > > The worst-case trade-off under this invariant is bounded > > > > > > > re-processing > > > > > > > proportional to the replication lag at failover time not total > > > > > partition > > > > > > > depth which is perfectly consistent with Kafka's documented > > > > > at-least-once > > > > > > > delivery guarantees. Broad enterprise production evidence from > > > > > large-scale > > > > > > > cross-cluster failovers confirms that state checks alone > > > > > > > (preventing > > > > > active > > > > > > > group overwrites) are insufficient; strict offset bounds clamping > > > > > > > is > > > > > > > required to achieve enterprise-grade data integrity. > > > > > > > > > > > > > > I look forward to your thoughts on implementing this final offset > > > > > capping > > > > > > > logic. Once VK12 is patched, I believe this architecture will be > > > > > > > exceptionally robust and ready for enterprise deployment. > > > > > > > > > > > > > > Best Regards, > > > > > > > > > > > > > > Viquar Khan > > > > > > > > > > > > > > > > > > > > > On Mon, 18 May 2026 at 14:23, Rajini Sivaram > > > > > > > <[email protected]> > > > > > > > wrote: > > > > > > > > > > > > > > > Hi Federico, > > > > > > > > > > > > > > > > Thanks for the updates! The KIP is looking good. A few more > > > > > > > > small > > > > > > > comments. > > > > > > > > > > > > > > > > > > > > > > > > RS13: A couple of places still refer to `kafka-mirror.sh` like > > > > > > > > under > > > > > > > > `Failover Process`. Can we change them to > > > > > `*kafka-cluster-mirrors.sh*`? > > > > > > > > > > > > > > > > RS14: Should we change `--entity-type mirrors` for > > > > > > > > kafka-configs to > > > > > be ` > > > > > > > > --entity-type *cluster-mirrors*` to be consistent? Also, > > > > > > > > CLUSTER_MIRROR((byte) > > > > > > > > 64, "mirror"); could be `*cluster-mirror*`? > > > > > > > > > > > > > > > > RS15: It may be useful to rename `mirror.topic.num.partitions` > > > > > > > > and ` > > > > > > > > mirror.topic.replication.factor` since they are very similar to > > > > > > > > ` > > > > > > > > mirror.topic.properties.exclude`, but the `mirror.topic` prefix > > > > > refers to > > > > > > > > different topics (the internal topic for the first two and > > > > > > > > actual > > > > > mirror > > > > > > > > topics for the other one). > > > > > > > > > > > > > > > > RS16: ACL Sync: KIP says "Deletes ACLs that exist in > > > > > > > > destination but > > > > > not > > > > > > > in > > > > > > > > source using DeleteAcls request." > > > > > > > > What happens if someone creates an ACL on the destination to > > > > > > > > deny > > > > > > > > User:Alice access to all topics? > > > > > > > > > > > > > > > > 1. If that ACL also existed on the source cluster and then > > > > > > > > it was > > > > > > > > removed, will it get removed from the destination? > > > > > > > > 2. If that ACL never existed on the source cluster, will it > > > > > > > > get > > > > > > > removed > > > > > > > > from the destination? > > > > > > > > > > > > > > > > RS17: The table in the Source ACLs section says: > > > > > "DescribeClusterMirrors > > > > > > > > MC Read Cluster Log truncation" > > > > > > > > Should that be `ClusterMirror:Read` instead of `Cluster:Read`? > > > > > > > > > > > > > > > > Regards, > > > > > > > > > > > > > > > > Rajini > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > On Fri, May 15, 2026 at 8:41 AM Federico Valeri < > > > > > [email protected]> > > > > > > > > wrote: > > > > > > > > > > > > > > > > > Hi Rajini, we finally addressed the API and tool naming > > > > > refactoring as > > > > > > > > > you suggested in RS6. Please take a look when you have time. > > > > > Thanks. > > > > > > > > > > > > > > > > > > > > > > > > > > > On Mon, May 11, 2026 at 6:17 PM Federico Valeri < > > > > > [email protected]> > > > > > > > > > wrote: > > > > > > > > > > > > > > > > > > > > Hello all, I want to highlight a couple of new paragraphs: > > > > > > > > > > > > > > > > > > > > 1. Leader Epoch Invariant: Cluster mirroring enforces the > > > > > invariant > > > > > > > > > > that the destination leader epoch must always be greater > > > > > > > > > > than or > > > > > > > equal > > > > > > > > > > to the source leader epoch (DLE>=SLE). Without this, > > > > > > > > > > consumers > > > > > on the > > > > > > > > > > destination cluster can get stuck in an infinite metadata > > > > > > > > > > refresh > > > > > > > loop > > > > > > > > > > when they encounter committed offsets carrying source epochs > > > > > higher > > > > > > > > > > than the local epoch. The invariant is maintained through > > > > > > > > > > three > > > > > > > > > > mechanisms: reactive bumping (epoch fencing triggered when > > > > > > > > > > SLE > > > > > > DLE > > > > > > > > > > during fetch), proactive bumping (scheduled when SLE > > > > > > > > > > approaches > > > > > DLE > > > > > > > > > > within a threshold), and periodic bumping (checked during > > > > > coordinator > > > > > > > > > > metadata sync). > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > https://cwiki.apache.org/confluence/pages/viewpage.action?pageId=406620973#KIP1279:ClusterMirroring-LeaderEpochInvariant > > > > > > > > > > > > > > > > > > > > 2. Group Offsets: The coordinator periodically syncs > > > > > > > > > > consumer and > > > > > > > > > > share group offsets from the source cluster to the > > > > > > > > > > destination > > > > > for > > > > > > > all > > > > > > > > > > mirrored topics. Groups are filtered by configurable > > > > > include/exclude > > > > > > > > > > patterns, and offsets are only synced for groups that are > > > > > > > > > > not > > > > > > > > > > currently active on the destination cluster, preventing > > > > > overwrites of > > > > > > > > > > local consumer progress. Because source and destination > > > > > > > > > > share the > > > > > > > same > > > > > > > > > > topic offsets (no offset translation), synced offsets can > > > > > > > > > > be used > > > > > > > > > > directly without mapping. > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > https://cwiki.apache.org/confluence/pages/viewpage.action?pageId=406620973#KIP1279:ClusterMirroring-GroupOffsets > > > > > > > > > > > > > > > > > > > > These new paragraphs directly address some of your > > > > > > > > > > questions, > > > > > but let > > > > > > > > > > me list them here: > > > > > > > > > > > > > > > > > > > > JR2: Yes, we removed the incorrect phrase and added more > > > > > > > > > > details > > > > > to > > > > > > > > > > the paragraph. > > > > > > > > > > > > > > > > > > > > JR4: When source cluster topic has tiered storage enabled, > > > > > > > > > > CM > > > > > works > > > > > > > by > > > > > > > > > > mirroring remote and local log into destination cluster. > > > > > > > > > > When > > > > > > > > > > destination cluster topic has tiered storage enabled, CM > > > > > > > > > > fails in > > > > > > > > > > PREPARING state because the LME may be in remote storage, > > > > > > > > > > but > > > > > works > > > > > > > > > > fine if already MIRRORING because no truncation is needed. > > > > > > > > > > > > > > > > > > > > JR11: See "Leader Epoch Invariant" paragraph mentioned > > > > > > > > > > above. > > > > > > > > > > > > > > > > > > > > JR13: We can't support stateful Streams application because > > > > > > > > > > asynchronous replication cannot preserve the transactional > > > > > boundaries > > > > > > > > > > between input offset commits, state store mutations written > > > > > > > > > > to > > > > > > > > > > changelog topics, and intermediate records written to > > > > > > > > > > repartition > > > > > > > > > > topics. The synchronous extension of this design will be > > > > > > > > > > able to > > > > > > > > > > support them. Existing Features Integration paragraph > > > > > > > > > > updated. > > > > > > > > > > > > > > > > > > > > JR18: See "Group Offsets" paragraph mentioned above. > > > > > > > > > > > > > > > > > > > > IY1: See "Group Offsets" paragraph mentioned above. > > > > > > > > > > > > > > > > > > > > Thanks > > > > > > > > > > Fede > > > > > > > > > > > > > > > > > > > > On Mon, May 11, 2026 at 6:08 PM Federico Valeri < > > > > > > > [email protected]> > > > > > > > > > wrote: > > > > > > > > > > > > > > > > > > > > > > Hi Vaquar, > > > > > > > > > > > > > > > > > > > > > > VK4/VK8: We don't do PID mapping anymore. The KIP was > > > > > > > > > > > updated > > > > > some > > > > > > > > > > > time ago with the new approach based on the new PID reset > > > > > control > > > > > > > > > > > record. > > > > > > > > > > > > > > > > > > > > > > VK11: The transaction index is always built locally > > > > > > > > > > > during log > > > > > > > > append, > > > > > > > > > > > never copied. > > > > > > > > > > > > > > > > > > > > > > VK1: The 50,000 * 1MB = 50GB calculation misunderstands > > > > > > > > > > > the > > > > > fetch > > > > > > > > > > > model. Fetcher threads don't allocate one buffer per > > > > > > > > > > > partition. > > > > > > > > Actual > > > > > > > > > > > peak memory is roughly num_fetcher_threads * > > > > > response_max_bytes, > > > > > > > not > > > > > > > > > > > num_partitions * partition_max_bytes. With 1 fetcher > > > > > > > > > > > thread > > > > > and the > > > > > > > > > > > default response max, the memory footprint is modest > > > > > regardless of > > > > > > > > > > > partition count. We are leveraging the same proven pattern > > > > > used by > > > > > > > > the > > > > > > > > > > > internal replication. > > > > > > > > > > > > > > > > > > > > > > VK3: The __mirror_state topic uses hash-based partitioning > > > > > based on > > > > > > > > > > > mirrorName, topicId and partition number. With the > > > > > > > > > > > production > > > > > > > default > > > > > > > > > > > of 50 partitions, 50,000 partition transitions distribute > > > > > across > > > > > > > ~50 > > > > > > > > > > > partition leaders on different brokers, not a single > > > > > > > > > > > broker. > > > > > This > > > > > > > is > > > > > > > > > > > the same proven pattern as __consumer_offsets, which > > > > > > > > > > > handles > > > > > > > millions > > > > > > > > > > > of commits. > > > > > > > > > > > > > > > > > > > > > > VK12: The scenario described can only occur if offsets are > > > > > > > > > > > force-written to an active group, which the design > > > > > > > > > > > prevents. > > > > > > > > > > > > > > > > > > > > > > Cheers > > > > > > > > > > > Fede > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > >
