artemlivshits commented on code in PR #13391: URL: https://github.com/apache/kafka/pull/13391#discussion_r1148060963
########## core/src/main/scala/kafka/server/AddPartitionsToTxnManager.scala: ########## @@ -0,0 +1,173 @@ +/** + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package kafka.server + +import kafka.common.{InterBrokerSendThread, RequestAndCompletionHandler} +import org.apache.kafka.clients.{ClientResponse, NetworkClient, RequestCompletionHandler} +import org.apache.kafka.common.{InvalidRecordException, Node, TopicPartition} +import org.apache.kafka.common.message.AddPartitionsToTxnRequestData.{AddPartitionsToTxnTransaction, AddPartitionsToTxnTransactionCollection} +import org.apache.kafka.common.protocol.Errors +import org.apache.kafka.common.requests.{AddPartitionsToTxnRequest, AddPartitionsToTxnResponse} +import org.apache.kafka.common.utils.Time + +import java.util.Collections +import scala.collection.mutable + +object AddPartitionsToTxnManager { + type AppendCallback = Map[TopicPartition, Errors] => Unit +} + + +class TransactionDataAndCallbacks(val transactionData: AddPartitionsToTxnTransactionCollection, + val callbacks: mutable.Map[String, AddPartitionsToTxnManager.AppendCallback]) + + +class AddPartitionsToTxnManager(config: KafkaConfig, client: NetworkClient, time: Time) + extends InterBrokerSendThread("AddPartitionsToTxnSenderThread-" + config.brokerId, client, config.requestTimeoutMs, time) { + + private val inflightNodes = mutable.HashSet[Node]() + private val nodesToTransactions = mutable.Map[Node, TransactionDataAndCallbacks]() + + def addTxnData(node: Node, transactionData: AddPartitionsToTxnTransaction, callback: AddPartitionsToTxnManager.AppendCallback): Unit = { + // Check if we have already (either node or individual transaction). + val currentNodeAndTransactionDataOpt = nodesToTransactions.get(node) + currentNodeAndTransactionDataOpt match { + case None => + nodesToTransactions.put(node, + new TransactionDataAndCallbacks(new AddPartitionsToTxnTransactionCollection(Collections.singletonList(transactionData).iterator()), + mutable.Map(transactionData.transactionalId() -> callback))) + case Some(currentNodeAndTransactionData) => + // Check if we already have txn ID -- this should only happen in epoch bump case. If so, we should return error for old entry and remove from queue. + val currentTransactionData = currentNodeAndTransactionData.transactionData.find(transactionData.transactionalId) + if (currentTransactionData != null) { + if (currentTransactionData.producerEpoch() < transactionData.producerEpoch()) { + val topicPartitionsToError = mutable.Map[TopicPartition, Errors]() + currentTransactionData.topics().forEach { topic => + topic.partitions().forEach { partition => + topicPartitionsToError.put(new TopicPartition(topic.name(), partition), Errors.INVALID_PRODUCER_EPOCH) + } + } + val oldCallback = currentNodeAndTransactionData.callbacks(transactionData.transactionalId()) + currentNodeAndTransactionData.transactionData.remove(transactionData) + oldCallback(topicPartitionsToError.toMap) + } else { + // We should never see a request on the same epoch since we haven't finished handling the one in queue + throw new InvalidRecordException("Received a second request from the same connection without finishing the first.") + } + } + currentNodeAndTransactionData.transactionData.add(transactionData) + currentNodeAndTransactionData.callbacks.put(transactionData.transactionalId(), callback) + } + wakeup() + } + + private class AddPartitionsToTxnHandler(node: Node, transactionDataAndCallbacks: TransactionDataAndCallbacks) extends RequestCompletionHandler { + override def onComplete(response: ClientResponse): Unit = { Review Comment: We should measure the time it took to process the request to TC and report it (either as "remote time" that we already have or as a new metric), otherwise it'll just roll into "local time" (time spent by this broker to handle the request). ########## core/src/main/scala/kafka/server/AddPartitionsToTxnManager.scala: ########## @@ -0,0 +1,173 @@ +/** + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package kafka.server + +import kafka.common.{InterBrokerSendThread, RequestAndCompletionHandler} +import org.apache.kafka.clients.{ClientResponse, NetworkClient, RequestCompletionHandler} +import org.apache.kafka.common.{InvalidRecordException, Node, TopicPartition} +import org.apache.kafka.common.message.AddPartitionsToTxnRequestData.{AddPartitionsToTxnTransaction, AddPartitionsToTxnTransactionCollection} +import org.apache.kafka.common.protocol.Errors +import org.apache.kafka.common.requests.{AddPartitionsToTxnRequest, AddPartitionsToTxnResponse} +import org.apache.kafka.common.utils.Time + +import java.util.Collections +import scala.collection.mutable + +object AddPartitionsToTxnManager { + type AppendCallback = Map[TopicPartition, Errors] => Unit +} + + +class TransactionDataAndCallbacks(val transactionData: AddPartitionsToTxnTransactionCollection, + val callbacks: mutable.Map[String, AddPartitionsToTxnManager.AppendCallback]) + + +class AddPartitionsToTxnManager(config: KafkaConfig, client: NetworkClient, time: Time) + extends InterBrokerSendThread("AddPartitionsToTxnSenderThread-" + config.brokerId, client, config.requestTimeoutMs, time) { + + private val inflightNodes = mutable.HashSet[Node]() + private val nodesToTransactions = mutable.Map[Node, TransactionDataAndCallbacks]() + + def addTxnData(node: Node, transactionData: AddPartitionsToTxnTransaction, callback: AddPartitionsToTxnManager.AppendCallback): Unit = { + // Check if we have already (either node or individual transaction). + val currentNodeAndTransactionDataOpt = nodesToTransactions.get(node) + currentNodeAndTransactionDataOpt match { + case None => + nodesToTransactions.put(node, + new TransactionDataAndCallbacks(new AddPartitionsToTxnTransactionCollection(Collections.singletonList(transactionData).iterator()), + mutable.Map(transactionData.transactionalId() -> callback))) + case Some(currentNodeAndTransactionData) => + // Check if we already have txn ID -- this should only happen in epoch bump case. If so, we should return error for old entry and remove from queue. + val currentTransactionData = currentNodeAndTransactionData.transactionData.find(transactionData.transactionalId) + if (currentTransactionData != null) { + if (currentTransactionData.producerEpoch() < transactionData.producerEpoch()) { + val topicPartitionsToError = mutable.Map[TopicPartition, Errors]() + currentTransactionData.topics().forEach { topic => + topic.partitions().forEach { partition => + topicPartitionsToError.put(new TopicPartition(topic.name(), partition), Errors.INVALID_PRODUCER_EPOCH) + } + } + val oldCallback = currentNodeAndTransactionData.callbacks(transactionData.transactionalId()) + currentNodeAndTransactionData.transactionData.remove(transactionData) + oldCallback(topicPartitionsToError.toMap) + } else { + // We should never see a request on the same epoch since we haven't finished handling the one in queue Review Comment: If a request hits a timeout the new request will come in the new connection, so the fact the old connection is muted wouldn't prevent the new request to come. The timeout processing on the client is just a timer, once it expires, it'll kill old connection, create a new one and re-send the batch. We won't bump the epoch on retry -- it'll override duplicate checking logic, so we'd have duplicates if we did, so the exact same batch (same epoch, same sequence) will come again on the new connection. ########## core/src/main/scala/kafka/server/AddPartitionsToTxnManager.scala: ########## @@ -0,0 +1,173 @@ +/** + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package kafka.server + +import kafka.common.{InterBrokerSendThread, RequestAndCompletionHandler} +import org.apache.kafka.clients.{ClientResponse, NetworkClient, RequestCompletionHandler} +import org.apache.kafka.common.{InvalidRecordException, Node, TopicPartition} +import org.apache.kafka.common.message.AddPartitionsToTxnRequestData.{AddPartitionsToTxnTransaction, AddPartitionsToTxnTransactionCollection} +import org.apache.kafka.common.protocol.Errors +import org.apache.kafka.common.requests.{AddPartitionsToTxnRequest, AddPartitionsToTxnResponse} +import org.apache.kafka.common.utils.Time + +import java.util.Collections +import scala.collection.mutable + +object AddPartitionsToTxnManager { + type AppendCallback = Map[TopicPartition, Errors] => Unit +} + + +class TransactionDataAndCallbacks(val transactionData: AddPartitionsToTxnTransactionCollection, + val callbacks: mutable.Map[String, AddPartitionsToTxnManager.AppendCallback]) + + +class AddPartitionsToTxnManager(config: KafkaConfig, client: NetworkClient, time: Time) + extends InterBrokerSendThread("AddPartitionsToTxnSenderThread-" + config.brokerId, client, config.requestTimeoutMs, time) { + + private val inflightNodes = mutable.HashSet[Node]() + private val nodesToTransactions = mutable.Map[Node, TransactionDataAndCallbacks]() + + def addTxnData(node: Node, transactionData: AddPartitionsToTxnTransaction, callback: AddPartitionsToTxnManager.AppendCallback): Unit = { + // Check if we have already (either node or individual transaction). + val currentNodeAndTransactionDataOpt = nodesToTransactions.get(node) + currentNodeAndTransactionDataOpt match { + case None => + nodesToTransactions.put(node, + new TransactionDataAndCallbacks(new AddPartitionsToTxnTransactionCollection(Collections.singletonList(transactionData).iterator()), + mutable.Map(transactionData.transactionalId() -> callback))) + case Some(currentNodeAndTransactionData) => + // Check if we already have txn ID -- this should only happen in epoch bump case. If so, we should return error for old entry and remove from queue. + val currentTransactionData = currentNodeAndTransactionData.transactionData.find(transactionData.transactionalId) + if (currentTransactionData != null) { + if (currentTransactionData.producerEpoch() < transactionData.producerEpoch()) { Review Comment: Oh, I see, we have only one request for a trasnactional.id, so we need to complete previous one in order to replace. >The only time we can receive two requests from the same txn id is when the producer restarts and the epoch is bumped See my other comments about retries. ########## core/src/main/scala/kafka/server/AddPartitionsToTxnManager.scala: ########## @@ -0,0 +1,173 @@ +/** + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package kafka.server + +import kafka.common.{InterBrokerSendThread, RequestAndCompletionHandler} +import org.apache.kafka.clients.{ClientResponse, NetworkClient, RequestCompletionHandler} +import org.apache.kafka.common.{InvalidRecordException, Node, TopicPartition} +import org.apache.kafka.common.message.AddPartitionsToTxnRequestData.{AddPartitionsToTxnTransaction, AddPartitionsToTxnTransactionCollection} +import org.apache.kafka.common.protocol.Errors +import org.apache.kafka.common.requests.{AddPartitionsToTxnRequest, AddPartitionsToTxnResponse} +import org.apache.kafka.common.utils.Time + +import java.util.Collections +import scala.collection.mutable + +object AddPartitionsToTxnManager { + type AppendCallback = Map[TopicPartition, Errors] => Unit +} + + +class TransactionDataAndCallbacks(val transactionData: AddPartitionsToTxnTransactionCollection, + val callbacks: mutable.Map[String, AddPartitionsToTxnManager.AppendCallback]) + + +class AddPartitionsToTxnManager(config: KafkaConfig, client: NetworkClient, time: Time) + extends InterBrokerSendThread("AddPartitionsToTxnSenderThread-" + config.brokerId, client, config.requestTimeoutMs, time) { + + private val inflightNodes = mutable.HashSet[Node]() + private val nodesToTransactions = mutable.Map[Node, TransactionDataAndCallbacks]() + + def addTxnData(node: Node, transactionData: AddPartitionsToTxnTransaction, callback: AddPartitionsToTxnManager.AppendCallback): Unit = { + // Check if we have already (either node or individual transaction). + val currentNodeAndTransactionDataOpt = nodesToTransactions.get(node) + currentNodeAndTransactionDataOpt match { + case None => + nodesToTransactions.put(node, + new TransactionDataAndCallbacks(new AddPartitionsToTxnTransactionCollection(Collections.singletonList(transactionData).iterator()), + mutable.Map(transactionData.transactionalId() -> callback))) + case Some(currentNodeAndTransactionData) => + // Check if we already have txn ID -- this should only happen in epoch bump case. If so, we should return error for old entry and remove from queue. + val currentTransactionData = currentNodeAndTransactionData.transactionData.find(transactionData.transactionalId) + if (currentTransactionData != null) { + if (currentTransactionData.producerEpoch() < transactionData.producerEpoch()) { + val topicPartitionsToError = mutable.Map[TopicPartition, Errors]() + currentTransactionData.topics().forEach { topic => + topic.partitions().forEach { partition => + topicPartitionsToError.put(new TopicPartition(topic.name(), partition), Errors.INVALID_PRODUCER_EPOCH) + } + } + val oldCallback = currentNodeAndTransactionData.callbacks(transactionData.transactionalId()) + currentNodeAndTransactionData.transactionData.remove(transactionData) + oldCallback(topicPartitionsToError.toMap) + } else { + // We should never see a request on the same epoch since we haven't finished handling the one in queue + throw new InvalidRecordException("Received a second request from the same connection without finishing the first.") + } + } + currentNodeAndTransactionData.transactionData.add(transactionData) + currentNodeAndTransactionData.callbacks.put(transactionData.transactionalId(), callback) + } + wakeup() + } + + private class AddPartitionsToTxnHandler(node: Node, transactionDataAndCallbacks: TransactionDataAndCallbacks) extends RequestCompletionHandler { + override def onComplete(response: ClientResponse): Unit = { + inflightNodes.synchronized(inflightNodes.remove(node)) Review Comment: I'd expect the synchronization would be added as part of making the class properly multithreaded. ########## core/src/main/scala/kafka/network/RequestChannel.scala: ########## @@ -483,14 +493,10 @@ class RequestChannel(val queueSize: Int, def sendShutdownRequest(): Unit = requestQueue.put(ShutdownRequest) - // TODO: this is the most straightforward implementation, we may want to address fairness: - // The callback function is invoked as part of processing a request, so it already "stood in line" - // when the original request got accepted, so it may be unfair to make put to the end of the line. - // A simple solution would be to put it to the head of the line, but then multiple callbacks would - // be in LIFO order, which is weird. - // We may want to use a PriorityBlockingQueue or just have 2 queues and write customer wait / notify - // synchronization. - def sendCallbackRequest(request: CallbackRequest): Unit = requestQueue.put(request) + def sendCallbackRequest(request: CallbackRequest): Unit = { + callbackQueue.put(request) Review Comment: We could try to re-purpose the action queue for this, but it currently has different semantics -- the error processing is different, for example (it's executed in the finally clause). What we want here is the semantics that we'd get if we just blocked the thread while we're contacting TC, but without blocking the thread. Same error handling, same context, same metrics, etc. The basic implementation of the desired semantics is just a queue, but we could also preserve request context (see TODOs in KafkaRequestHandler.scala) so that this functionality that would work for any requests that need to do non-blocking async calls. ########## core/src/main/scala/kafka/network/RequestChannel.scala: ########## @@ -481,6 +493,11 @@ class RequestChannel(val queueSize: Int, def sendShutdownRequest(): Unit = requestQueue.put(ShutdownRequest) + def sendCallbackRequest(request: CallbackRequest): Unit = { + callbackQueue.put(request) + requestQueue.put(RequestChannel.WakeupRequest) Review Comment: I guess, if the request queue is full then it means that the request threads are all overloaded, so if we block the inter-broker channel thread it probably wouldn't matter much as the broker is already in pain. But on the other hand, it's kind of strange to not implement technically more correct behavior because we've got some spurious warnings. ########## core/src/main/scala/kafka/server/AddPartitionsToTxnManager.scala: ########## @@ -0,0 +1,173 @@ +/** + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package kafka.server + +import kafka.common.{InterBrokerSendThread, RequestAndCompletionHandler} +import org.apache.kafka.clients.{ClientResponse, NetworkClient, RequestCompletionHandler} +import org.apache.kafka.common.{InvalidRecordException, Node, TopicPartition} +import org.apache.kafka.common.message.AddPartitionsToTxnRequestData.{AddPartitionsToTxnTransaction, AddPartitionsToTxnTransactionCollection} +import org.apache.kafka.common.protocol.Errors +import org.apache.kafka.common.requests.{AddPartitionsToTxnRequest, AddPartitionsToTxnResponse} +import org.apache.kafka.common.utils.Time + +import java.util.Collections +import scala.collection.mutable + +object AddPartitionsToTxnManager { + type AppendCallback = Map[TopicPartition, Errors] => Unit +} + + +class TransactionDataAndCallbacks(val transactionData: AddPartitionsToTxnTransactionCollection, + val callbacks: mutable.Map[String, AddPartitionsToTxnManager.AppendCallback]) + + +class AddPartitionsToTxnManager(config: KafkaConfig, client: NetworkClient, time: Time) + extends InterBrokerSendThread("AddPartitionsToTxnSenderThread-" + config.brokerId, client, config.requestTimeoutMs, time) { + + private val inflightNodes = mutable.HashSet[Node]() + private val nodesToTransactions = mutable.Map[Node, TransactionDataAndCallbacks]() + + def addTxnData(node: Node, transactionData: AddPartitionsToTxnTransaction, callback: AddPartitionsToTxnManager.AppendCallback): Unit = { + // Check if we have already (either node or individual transaction). + val currentNodeAndTransactionDataOpt = nodesToTransactions.get(node) + currentNodeAndTransactionDataOpt match { + case None => + nodesToTransactions.put(node, + new TransactionDataAndCallbacks(new AddPartitionsToTxnTransactionCollection(Collections.singletonList(transactionData).iterator()), + mutable.Map(transactionData.transactionalId() -> callback))) + case Some(currentNodeAndTransactionData) => + // Check if we already have txn ID -- this should only happen in epoch bump case. If so, we should return error for old entry and remove from queue. + val currentTransactionData = currentNodeAndTransactionData.transactionData.find(transactionData.transactionalId) + if (currentTransactionData != null) { + if (currentTransactionData.producerEpoch() < transactionData.producerEpoch()) { + val topicPartitionsToError = mutable.Map[TopicPartition, Errors]() + currentTransactionData.topics().forEach { topic => + topic.partitions().forEach { partition => + topicPartitionsToError.put(new TopicPartition(topic.name(), partition), Errors.INVALID_PRODUCER_EPOCH) + } + } + val oldCallback = currentNodeAndTransactionData.callbacks(transactionData.transactionalId()) + currentNodeAndTransactionData.transactionData.remove(transactionData) + oldCallback(topicPartitionsToError.toMap) + } else { + // We should never see a request on the same epoch since we haven't finished handling the one in queue + throw new InvalidRecordException("Received a second request from the same connection without finishing the first.") + } + } + currentNodeAndTransactionData.transactionData.add(transactionData) + currentNodeAndTransactionData.callbacks.put(transactionData.transactionalId(), callback) + } + wakeup() + } + + private class AddPartitionsToTxnHandler(node: Node, transactionDataAndCallbacks: TransactionDataAndCallbacks) extends RequestCompletionHandler { + override def onComplete(response: ClientResponse): Unit = { + inflightNodes.synchronized(inflightNodes.remove(node)) + if (response.authenticationException() != null) { + error(s"AddPartitionsToTxnRequest failed for broker ${config.brokerId} with an " + + "authentication exception.", response.authenticationException) + transactionDataAndCallbacks.callbacks.foreach { case (txnId, callback) => + callback(buildErrorMap(txnId, transactionDataAndCallbacks.transactionData, Errors.forException(response.authenticationException()).code())) + } + } else if (response.versionMismatch != null) { + // We may see unsupported version exception if we try to send a verify only request to a broker that can't handle it. + // In this case, skip verification. + error(s"AddPartitionsToTxnRequest failed for broker ${config.brokerId} with invalid version exception. This suggests verification is not supported." + + s"Continuing handling the produce request.") + transactionDataAndCallbacks.callbacks.values.foreach(_(Map.empty)) + } else { + val addPartitionsToTxnResponseData = response.responseBody.asInstanceOf[AddPartitionsToTxnResponse].data + if (addPartitionsToTxnResponseData.errorCode != 0) { + error(s"AddPartitionsToTxnRequest for broker ${config.brokerId} returned with error ${Errors.forCode(addPartitionsToTxnResponseData.errorCode)}.") + // TODO: send error back correctly -- we need to verify all possible errors can be handled by the client. + // errors -- versionmismatch --> handled above + // -- clusterauth --> should handle differently + transactionDataAndCallbacks.callbacks.foreach { case (txnId, callback) => + callback(buildErrorMap(txnId, transactionDataAndCallbacks.transactionData, addPartitionsToTxnResponseData.errorCode())) + } + } else { + val unverified = mutable.Map[TopicPartition, Errors]() + addPartitionsToTxnResponseData.resultsByTransaction().forEach { transactionResult => + transactionResult.topicResults().forEach { topicResult => + topicResult.resultsByPartition().forEach { partitionResult => + val tp = new TopicPartition(topicResult.name(), partitionResult.partitionIndex()) + if (partitionResult.partitionErrorCode() != Errors.NONE.code()) { + // Producers expect to handle INVALID_PRODUCER_EPOCH in this scenario. + val code = + if (partitionResult.partitionErrorCode() == Errors.PRODUCER_FENCED.code()) + Errors.INVALID_PRODUCER_EPOCH.code() + else + partitionResult.partitionErrorCode() + unverified.put(tp, Errors.forCode(code)) + } + } + } + val callback = transactionDataAndCallbacks.callbacks(transactionResult.transactionalId()) + callback(unverified.toMap) + unverified.clear() Review Comment: Should we move it to the proper scope then? (the indentation is hard to follow in the PR view, I assumed it's already in the proper scope) -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: jira-unsubscr...@kafka.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
