Kafka

This repo is for solving kafka related problem

FETCH_SESSION_ID_NOT_FOUND

Error: org.apache.kafka.clients.FetchSessionHandler : [Consumer clientId=xxx, groupId=xxx] Node x was unable to process the fetch request with (sessionId=224593981, epoch=969): FETCH_SESSION_ID_NOT_FOUND.

Questions

1. What is a fetch request?
2. What is a fetch session ?
3. The number of session depend on what?
4. What is a fetch session id ?
5. What is fetch epoch ?
6. Who decide how long session is alive ? What conf ?
7. What is FETCH_SESSION_ID_NOT_FOUND
8. Tracking number of FETCH_SESSION_ID_NOT_FOUND
9. Best pratice to configure MAX_FETCH_SESSION

Answers

1. Fetch requests:

   Sent by consumers and follower replicas when they read messages from Kafka brokers.

   • Consumers send fetch requests to fetch data from topics
   • Follower replicas send fetch requests to keep in-sync with master node

2. Fetch session: A fetch session encapsulates the state of individual fetcher. This allows kafka to avoid resending this state as part of each fetch request. The fetch session includes:

   1. A randomly generated 32bit session id which is unique on the leader
   2. The 32bit fetch epoch
   3. Cached data about each partition which the fetchers is interested in
   4. The privileged bit: is set if the fetch session was created by a follower. It is cleared if the fetch session as created by a reguler consumer.
   5. The time when the fetch session was last used: This is the time in wall-clock milliseconds when the fetch session was last used. This is used to expire fetch sessions after they have been inactive. Cache session data about each partition:
   6. Topic name
   7. The partition index
   8. The maximum number of bytes to fetch from this partition
   9. The fetch offset
   10. The high water mark
   11. The fetcher log start offset
   12. The leader log start offset

   The leader uses this cached information to decide which partitions to include in FetchReponse. Whenever any of these element change, or if there is new data available for a partition, the partition will be include

3. The number of fetch session: As you see, the number of fetch sessions are depend on the number of regular consumers and the number of follower. nConsumers + nReplicas For example: With 3 brokers, 2k topics, 4k partitions(2 for each), 8k replicas(2 for each). The number of consumer should be less then the number of partitions. nSession <= 4k + 9k = 12k

4. Fetch session id: A randomly generated 32bit session id which is unique on the leader. It is unique, immutable identifier for each session. Since IDs is randomly generated, it cannot leak information to unpriveleged clients.

5. Fetch epoch: The fetch epoch is monotonically incrementing 32bit counter. After processing request N, the brokers expects to receive request N + 1. The sequence number is always greater then 0. After reaching MAX_INT, it wraps around to 1.

6. Who decide how long session is alive ? What conf ? Because fetch sessions use memory on the leader, we want to limit the amount of them that we have at any given time. Therefor, each broker will create only a limited number of incremental fetch sessions. max.incremental.fetch.session.cache.slots: which sets the number of incremental fetch session cache slots on the broker. Default value: 1000 min.incremental.fetch.session.eviction.ms: which sets the minumin amount of time we will wait before evicting an incremental fetch session from the cache. Value: 120000(ms) = 120(sec) = 2(min) When the servers gets a new request to create an incremental fetch session, it will compare the proposed new session with its existing sessions. The new session will evict an existing session if and only if: 1. The new session belongs to a follower, and the existing session belongs to a reguler consumer, OR 2. The existing session has been inactive for more than min.incremental.fetch.session.eviction.ms, OR 3. The existing session has existed for more than min.incremental.fetch.session.eviction.ms, AND the new session has more partitions.

7. What is FETCH_SESSION_ID_NOT_FOUND: The server responds with this error code when the client request refers to a fetch session that the server does not know about. This may happen if there was a client error, or if the fetch session was evicted by the server.

8. Tracking number of FETCH_SESSION_ID_NOT_FOUND: NumIncrementalFetchSessions: Tracks the number of incremental fetch sessions which exist NumIncrementalFetchPartitionsCached: Tracks the number of partitions cached by incremental fetch sessions.

###Refs https://stackoverflow.com/questions/58484342/kafka-continuously-getting-fetch-session-id-not-found