jeffkbkim commented on code in PR #13870:
URL: https://github.com/apache/kafka/pull/13870#discussion_r1235808438
##########
group-coordinator/src/main/java/org/apache/kafka/coordinator/group/GroupMetadataManager.java:
##########
@@ -874,4 +1045,1265 @@ public void replay(
consumerGroup.updateMember(newMember);
}
}
+
+ // Below stores all methods to handle generic group APIs.
+
+ /**
+ * Handle a JoinGroupRequest.
+ *
+ * @param context The request context.
+ * @param request The actual JoinGroup request.
+ *
+ * @return The result that contains records to append if the join group
phase completes.
+ */
+ public CoordinatorResult<CompletableFuture<Errors>, Record>
genericGroupJoin(
+ RequestContext context,
+ JoinGroupRequestData request,
+ CompletableFuture<JoinGroupResponseData> responseFuture
+ ) {
+ CoordinatorResult<CompletableFuture<Errors>, Record> result =
EMPTY_RESULT;
+ String groupId = request.groupId();
+ String memberId = request.memberId();
+ int sessionTimeoutMs = request.sessionTimeoutMs();
+
+ if (sessionTimeoutMs < groupMinSessionTimeoutMs ||
+ sessionTimeoutMs > groupMaxSessionTimeoutMs
+ ) {
+ responseFuture.complete(new JoinGroupResponseData()
+ .setMemberId(memberId)
+ .setErrorCode(Errors.INVALID_SESSION_TIMEOUT.code())
+ );
+ } else {
+ boolean isUnknownMember = memberId.equals(UNKNOWN_MEMBER_ID);
+ // Group is created if it does not exist and the member id is
UNKNOWN. if member
+ // is specified but group does not exist, request is rejected with
GROUP_ID_NOT_FOUND
+ GenericGroup group;
+ try {
+ group = (GenericGroup) getOrMaybeCreateGroup(groupId, GENERIC,
isUnknownMember);
+ } catch (Throwable t) {
+ responseFuture.complete(new JoinGroupResponseData()
+ .setMemberId(memberId)
+ .setErrorCode(Errors.forException(t).code())
+ );
+ return EMPTY_RESULT;
+ }
+
+ String joinReason = request.reason();
+ if (joinReason == null || joinReason.isEmpty()) {
+ joinReason = "not provided";
+ }
+
+ if (!acceptJoiningMember(group, memberId)) {
+ group.remove(memberId);
+ responseFuture.complete(new JoinGroupResponseData()
+ .setMemberId(UNKNOWN_MEMBER_ID)
+ .setErrorCode(Errors.GROUP_MAX_SIZE_REACHED.code())
+ );
+
+ } else if (isUnknownMember) {
+ result = genericGroupJoinNewMember(
+ context,
+ request,
+ group,
+ joinReason,
+ responseFuture
+ );
+ } else {
+ result = genericGroupJoinExistingMember(
+ context,
+ request,
+ group,
+ joinReason,
+ responseFuture
+ );
+ }
+
+ // Attempt to complete join group phase. We do not complete
+ // the join group phase if this is the initial rebalance.
+ if (group.isInState(PREPARING_REBALANCE) &&
+ group.hasAllMembersJoined() &&
+ group.generationId() != 0
+ ) {
+ completeGenericGroupJoin(group);
+ }
+ }
+
+ return result;
+ }
+
+ /**
+ * Handle a new member generic group join.
+ *
+ * @param context The request context.
+ * @param request The join group request.
+ * @param group The group to add the member.
+ * @param joinReason The client reason for the join request.
+ * @param responseFuture The response future to complete.
+ *
+ * @return The coordinator result that will be appended to the log.
+ */
+ private CoordinatorResult<CompletableFuture<Errors>, Record>
genericGroupJoinNewMember(
+ RequestContext context,
+ JoinGroupRequestData request,
+ GenericGroup group,
+ String joinReason,
+ CompletableFuture<JoinGroupResponseData> responseFuture
+ ) {
+ List<Protocol> protocols = new ArrayList<>();
+ request.protocols().forEach(protocol -> protocols.add(new
Protocol(protocol.name(), protocol.metadata())));
+ if (group.isInState(DEAD)) {
+ // if the group is marked as dead, it means some other thread has
just removed the group
+ // from the coordinator metadata; it is likely that the group has
migrated to some other
+ // coordinator OR the group is in a transient unstable phase. Let
the member retry
+ // finding the correct coordinator and rejoin.
+ responseFuture.complete(new JoinGroupResponseData()
+ .setMemberId(UNKNOWN_MEMBER_ID)
+ .setErrorCode(Errors.COORDINATOR_NOT_AVAILABLE.code())
+ );
+ } else if (!group.supportsProtocols(request.protocolType(),
GenericGroupMember.plainProtocolSet(protocols))) {
+ responseFuture.complete(new JoinGroupResponseData()
+ .setMemberId(UNKNOWN_MEMBER_ID)
+ .setErrorCode(Errors.INCONSISTENT_GROUP_PROTOCOL.code())
+ );
+ } else {
+ Optional<String> groupInstanceId =
Optional.ofNullable(request.groupInstanceId());
+ String newMemberId = group.generateMemberId(context.clientId(),
groupInstanceId);
+
+ if (groupInstanceId.isPresent()) {
+ return genericGroupJoinNewStaticMember(
+ context,
+ request,
+ group,
+ newMemberId,
+ joinReason,
+ responseFuture
+ );
+ } else {
+ return genericGroupJoinNewDynamicMember(
+ context,
+ request,
+ group,
+ newMemberId,
+ joinReason,
+ responseFuture
+ );
+ }
+ }
+
+ return EMPTY_RESULT;
+ }
+
+ /**
+ * Handle new static member join. If there was an existing member id for
the group instance id,
+ * replace that member. Otherwise, add the member and rebalance.
+ *
+ * @param context The request context.
+ * @param request The join group request.
+ * @param group The group to add the member.
+ * @param newMemberId The newly generated member id.
+ * @param joinReason The client reason for the join request.
+ * @param responseFuture The response future to complete.
+ *
+ * @return The coordinator result that will be appended to the log.
+ */
+ private CoordinatorResult<CompletableFuture<Errors>, Record>
genericGroupJoinNewStaticMember(
+ RequestContext context,
+ JoinGroupRequestData request,
+ GenericGroup group,
+ String newMemberId,
+ String joinReason,
+ CompletableFuture<JoinGroupResponseData> responseFuture
+ ) {
+ String groupInstanceId = request.groupInstanceId();
+ String existingMemberId = group.staticMemberId(groupInstanceId);
+ if (existingMemberId != null) {
+ log.info("Static member with groupInstanceId={} and unknown member
id joins " +
+ "group {} in {} state. Replacing previously mapped member
{} with this groupInstanceId.",
+ groupInstanceId, group.groupId(), group.currentState(),
existingMemberId);
+
+ return updateStaticMemberAndRebalance(
+ context,
+ request,
+ group,
+ existingMemberId,
+ newMemberId,
+ joinReason,
+ responseFuture
+ );
+
+ } else {
+ log.info("Static member with groupInstanceId={} and unknown member
id joins " +
+ "group {} in {} state. Created a new member id {} for this
member and added to the group.",
+ groupInstanceId, group.groupId(), group.currentState(),
newMemberId);
+
+ return addMemberAndRebalance(context, request, group, newMemberId,
joinReason, responseFuture);
+ }
+ }
+
+ /**
+ * Handle a new dynamic member join. If the member id field is required,
the group metadata manager
+ * will add the new member id to the pending members and respond with
MEMBER_ID_REQUIRED along with
+ * the new member id for the client to join with.
+ *
+ * Otherwise, add the new member to the group and rebalance.
+ *
+ * @param context The request context.
+ * @param request The join group request.
+ * @param group The group to add the member.
+ * @param newMemberId The newly generated member id.
+ * @param joinReason The client reason for the join request.
+ * @param responseFuture The response future to complete.
+ *
+ * @return The coordinator result that will be appended to the log.
+ */
+ private CoordinatorResult<CompletableFuture<Errors>, Record>
genericGroupJoinNewDynamicMember(
+ RequestContext context,
+ JoinGroupRequestData request,
+ GenericGroup group,
+ String newMemberId,
+ String joinReason,
+ CompletableFuture<JoinGroupResponseData> responseFuture
+ ) {
+ boolean requireKnownMemberId = context.apiVersion() >= 4
+ && (request.groupInstanceId() == null ||
request.groupInstanceId().isEmpty());
+
+ if (requireKnownMemberId) {
+ // If member id required, register the member in the pending
member list and send
+ // back a response to call for another join group request with
allocated member id.
+ log.info("Dynamic member with unknown member id joins group {} in
{} state. " +
+ "Created a new member id {} and requesting the member to
rejoin with this id.",
+ group.groupId(), group.currentState(), newMemberId);
+
+ group.addPendingMember(newMemberId);
+ String heartbeatKey = heartbeatKey(group.groupId(), newMemberId);
+
+ timer.schedule(
+ heartbeatKey,
+ request.sessionTimeoutMs(),
+ TimeUnit.MILLISECONDS,
+ () -> expireGenericGroupMemberHeartbeat(group, newMemberId)
+ );
+
+ responseFuture.complete(new JoinGroupResponseData()
+ .setMemberId(newMemberId)
+ .setErrorCode(Errors.MEMBER_ID_REQUIRED.code())
+ );
+
+ } else {
+ log.info("Dynamic member with unknown member id joins group {} in
state {}. " +
+ "Created a new member id {} and added the member to the
group.",
+ group.groupId(), group.currentState(), newMemberId);
+
+ return addMemberAndRebalance(context, request, group, newMemberId,
joinReason, responseFuture);
+ }
+
+ return EMPTY_RESULT;
+ }
+
+ /**
+ * Handle a join group request for an existing member.
+ *
+ * @param context The request context.
+ * @param request The join group request.
+ * @param group The group to add the member.
+ * @param joinReason The client reason for the join request.
+ * @param responseFuture The response future to complete.
+ *
+ * @return The coordinator result that will be appended to the log.
+ */
+ private CoordinatorResult<CompletableFuture<Errors>, Record>
genericGroupJoinExistingMember(
+ RequestContext context,
+ JoinGroupRequestData request,
+ GenericGroup group,
+ String joinReason,
+ CompletableFuture<JoinGroupResponseData> responseFuture
+ ) {
+ List<Protocol> protocols = request.protocols().stream()
+ .map(protocol -> new Protocol(protocol.name(),
protocol.metadata())).collect(Collectors.toList());
+
+ String memberId = request.memberId();
+ String groupInstanceId = request.groupInstanceId();
+
+ if (group.isInState(DEAD)) {
+ // If the group is marked as dead, it means the group was recently
removed the group
+ // from the coordinator metadata; it is likely that the group has
migrated to some other
+ // coordinator OR the group is in a transient unstable phase. Let
the member retry
+ // finding the correct coordinator and rejoin.
+ responseFuture.complete(new JoinGroupResponseData()
+ .setMemberId(memberId)
+ .setErrorCode(Errors.COORDINATOR_NOT_AVAILABLE.code())
+ );
+ } else if (!group.supportsProtocols(
+ request.protocolType(),
+ GenericGroupMember.plainProtocolSet(protocols))
+ ) {
+ responseFuture.complete(new JoinGroupResponseData()
+ .setMemberId(memberId)
+ .setErrorCode(Errors.INCONSISTENT_GROUP_PROTOCOL.code())
+ );
+ } else if (group.isPendingMember(memberId)) {
+ // A rejoining pending member will be accepted. Note that pending
member cannot be a static member.
+ if (groupInstanceId != null) {
+ throw new IllegalStateException("Received unexpected JoinGroup
with groupInstanceId=" +
+ groupInstanceId + " for pending member with memberId=" +
memberId);
+ }
+
+ log.debug("Pending dynamic member with id {} joins group {} in {}
state. Adding to the group now.",
+ memberId, group.groupId(), group.currentState());
+
+ return addMemberAndRebalance(
+ context,
+ request,
+ group,
+ memberId,
+ joinReason,
+ responseFuture
+ );
+ } else {
+ Optional<Errors> memberError = validateExistingMember(
+ group,
+ memberId,
+ groupInstanceId,
+ "join-group"
+ );
+
+ if (memberError.isPresent()) {
+ responseFuture.complete(new JoinGroupResponseData()
+ .setMemberId(memberId)
+ .setErrorCode(memberError.get().code())
+ );
+ } else {
+ GenericGroupMember member = group.member(memberId);
+ if (group.isInState(PREPARING_REBALANCE)) {
+ return updateMemberAndRebalance(
+ request,
+ group,
+ member,
+ "Member " + member.memberId() + " is joining group
during " +
+ group.stateAsString() + "; client reason: " +
joinReason,
+ responseFuture
+ );
+ } else if (group.isInState(COMPLETING_REBALANCE)) {
+ if (member.matches(protocols)) {
+ // Member is joining with the same metadata (which
could be because it failed to
+ // receive the initial JoinGroup response), so just
return current group information
+ // for the current generation.
+ List<JoinGroupResponseData.JoinGroupResponseMember>
members = Collections.emptyList();
+ if (group.isLeader(memberId)) {
+ members = group.currentGenericGroupMembers();
+ }
+ responseFuture.complete(new JoinGroupResponseData()
+ .setMembers(members)
+ .setMemberId(memberId)
+ .setGenerationId(group.generationId())
+ .setProtocolName(group.protocolName().orElse(null))
+ .setProtocolType(group.protocolType().orElse(null))
+ .setLeader(group.leaderOrNull())
+ .setSkipAssignment(false)
+ .setErrorCode(Errors.NONE.code())
+ );
+ } else {
+ // Member has changed metadata, so force a rebalance
+ return updateMemberAndRebalance(
+ request,
+ group,
+ member,
+ "Updating metadata for member " + memberId + "
during " +
+ group.stateAsString() + "; client reason: " +
joinReason,
+ responseFuture
+ );
+ }
+ } else if (group.isInState(STABLE)) {
+ if (group.isLeader(memberId)) {
+ // Force a rebalance if the leader sends JoinGroup;
+ // This allows the leader to trigger rebalances for
changes affecting assignment
+ // which do not affect the member metadata (such as
topic metadata changes for the consumer)
+ return updateMemberAndRebalance(
+ request,
+ group,
+ member,
+ "Leader " + memberId + " re-joining group during "
+
+ group.stateAsString() + "; client reason: " +
joinReason,
+ responseFuture
+ );
+ } else if (!member.matches(protocols)) {
+ return updateMemberAndRebalance(
+ request,
+ group,
+ member,
+ "Updating metadata for member " + memberId + "
during " +
+ group.stateAsString() + "; client reason: " +
joinReason,
+ responseFuture
+ );
+ } else {
+ // For followers with no actual change to their
metadata, just return group information
+ // for the current generation which will allow them to
issue SyncGroup.
+ responseFuture.complete(new JoinGroupResponseData()
+ .setMembers(Collections.emptyList())
+ .setMemberId(memberId)
+ .setGenerationId(group.generationId())
+ .setProtocolName(group.protocolName().orElse(null))
+ .setProtocolType(group.protocolType().orElse(null))
+ .setLeader(group.leaderOrNull())
+ .setSkipAssignment(false)
+ .setErrorCode(Errors.NONE.code())
+ );
+ }
+ } else {
+ // Group reached unexpected (Empty) state. Let the joining
member reset their generation and rejoin.
+ log.warn("Attempt to add rejoining member {} of group {}
in unexpected group state {}",
+ memberId, group.groupId(), group.stateAsString());
+
+ responseFuture.complete(new JoinGroupResponseData()
+ .setMemberId(memberId)
+ .setErrorCode(Errors.UNKNOWN_MEMBER_ID.code())
+ );
+ }
+ }
+ }
+
+ return EMPTY_RESULT;
+ }
+
+ /**
+ * Complete the join group phase. Remove all dynamic members that have not
rejoined
+ * during this stage and proceed with the next generation for this group.
The generation id
+ * is incremented and the group transitions to CompletingRebalance state
if there is at least
+ * one member.
+ *
+ * If the group is in Empty state, append a new group metadata record to
the log. Otherwise,
+ * complete all members' join group response futures and wait for sync
requests from members.
+ *
+ * @param group The group that is completing the join group phase.
+ *
+ * @return The coordinator result that will be appended to the log.
+ */
+ private CoordinatorResult<CompletableFuture<Errors>, Record>
completeGenericGroupJoin(
+ GenericGroup group
+ ) {
+ timer.cancel(joinKey(group.groupId()));
+ String groupId = group.groupId();
+
+ Map<String, GenericGroupMember> notYetRejoinedDynamicMembers =
+ group.notYetRejoinedMembers().entrySet().stream()
+ .filter(entry -> !entry.getValue().isStaticMember())
+ .collect(Collectors.toMap(Map.Entry::getKey,
Map.Entry::getValue));
+
+ if (!notYetRejoinedDynamicMembers.isEmpty()) {
+ notYetRejoinedDynamicMembers.values().forEach(failedMember -> {
+ group.remove(failedMember.memberId());
+ timer.cancel(heartbeatKey(group.groupId(),
failedMember.memberId()));
+ });
+
+ log.info("Group {} removed dynamic members who haven't joined: {}",
+ groupId, notYetRejoinedDynamicMembers.keySet());
+ }
+
+ if (group.isInState(DEAD)) {
+ log.info("Group {} is dead, skipping rebalance stage.", groupId);
+ } else if (!group.maybeElectNewJoinedLeader() &&
!group.allMembers().isEmpty()) {
+ // If all members are not rejoining, we will postpone the
completion
+ // of rebalance preparing stage, and send out another delayed
operation
+ // until session timeout removes all the non-responsive members.
+ log.error("Group {} could not complete rebalance because no
members rejoined.", groupId);
+
+ return tryCompleteJoinElseSchedule(group);
+ } else {
+ group.initNextGeneration();
+ if (group.isInState(EMPTY)) {
+ log.info("Group {} with generation {} is now empty ({}-{})",
+ groupId, group.generationId(), topicPartition.topic(),
topicPartition.partition());
+
+ CompletableFuture<Errors> appendFuture = new
CompletableFuture<>();
+ appendFuture.whenComplete((error, t) -> {
+ if (t != null || error != Errors.NONE) {
+ // We failed to write the empty group metadata. If the
broker fails before another rebalance,
+ // the previous generation written to the log will
become active again (and most likely timeout).
+ // This should be safe since there are no active
members in an empty generation, so we just warn.
+ String message = t != null ? t.getMessage() :
error.message();
+ log.warn("Failed to write empty metadata for group {}:
{}", group.groupId(), message);
+ }
+ });
+
+ List<Record> records = Collections.singletonList(
+ RecordHelpers.newGroupMetadataRecord(group,
metadataImage.features().metadataVersion())
+ );
+
+ return new CoordinatorResult<>(records, appendFuture);
+
+ } else {
+ log.info("Stabilized group {} generation {} ({}-{}) with {}
members",
+ groupId, group.generationId(), topicPartition.topic(),
+ topicPartition.partition(), group.size());
+
+ // Complete the awaiting join group response future for all
the members after rebalancing
+ group.allMembers().forEach(member -> {
+
+ List<JoinGroupResponseData.JoinGroupResponseMember>
members = Collections.emptyList();
+ if (group.isLeader(member.memberId())) {
+ members = group.currentGenericGroupMembers();
+ }
+
+ JoinGroupResponseData response = new
JoinGroupResponseData()
+ .setMembers(members)
+ .setMemberId(member.memberId())
+ .setGenerationId(group.generationId())
+ .setProtocolName(group.protocolName().orElse(null))
+ .setProtocolType(group.protocolType().orElse(null))
+ .setLeader(group.leaderOrNull())
+ .setSkipAssignment(false)
+ .setErrorCode(Errors.NONE.code());
+
+ group.completeJoinFuture(member, response);
+ rescheduleGenericGroupMemberHeartbeat(group, member);
+ member.setIsNew(false);
+
+ group.addPendingSyncMember(member.memberId());
+ });
+
+ schedulePendingSync(group);
+ }
+ }
+
+ return EMPTY_RESULT;
+ }
+
+ /**
+ * Wait for sync requests for the group.
+ *
+ * @param group The group.
+ */
+ private void schedulePendingSync(GenericGroup group) {
+ timer.schedule(syncKey(group.groupId()),
+ group.rebalanceTimeoutMs(),
+ TimeUnit.MILLISECONDS,
+ () -> expirePendingSync(group, group.generationId()));
+ }
+
+ /**
+ * Invoked when the heartbeat operation is expired from the timer.
Possibly remove the member and
+ * try complete the join phase.
+ *
+ * @param group The group.
+ * @param memberId The member id.
+ *
+ * @return The coordinator result that will be appended to the log.
+ */
+ private CoordinatorResult<CompletableFuture<Errors>, Record>
expireGenericGroupMemberHeartbeat(
+ GenericGroup group,
+ String memberId
+ ) {
+ if (group.isInState(DEAD)) {
+ log.info("Received notification of heartbeat expiration for member
{} after group {} " +
+ "had already been unloaded or deleted.",
+ memberId, group.groupId());
+
+ } else if (group.isPendingMember(memberId)) {
+ log.info("Pending member {} in group {} has been removed after
session timeout expiration.",
+ memberId, group.groupId());
+
+ return removePendingMemberAndUpdateGenericGroup(group, memberId);
+ } else if (!group.hasMemberId(memberId)) {
+ log.debug("Member {} has already been removed from the group.",
memberId);
+ } else {
+ GenericGroupMember member = group.member(memberId);
+ if (!member.hasSatisfiedHeartbeat()) {
Review Comment:
if the member is awaiting on a join/sync response then we can't remove the
member on hb expiration right?
--
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
