Hi Oleg,
do you have configured your consumer/producer with "no data loss" configuration like bellow ? For Consumer, set auto.commit.enabled=false in consumer.properties For Producer 1. max.in.flight.requests.per.connection=1 2. retries=Int.MaxValue 3. acks=-1 4. block.on.buffer.full=true Like advised in this topic: https://community.hortonworks.com/articles/79891/kafka-mirror-maker-best-practices.html I have the following assumption: Your MM can't reach your targeted cluster (rc.exchange.jpy right ?) for this reason, your producer will retry indefinitely. But because mirror maker will block on producer buffer when is full, and now because your buffer is full and you have a retries policy + block to true combined with acks all, the wholeness of these parameter may exceed the retention time of your records. Do you have tracked this lead ? (It's just an idea) I hope you solve quickly your issue. Adrien ________________________________ De : Oleg Danilovich <oleg.danilov...@expcapital.com> Envoyé : samedi 3 mars 2018 15:42:47 À : users@kafka.apache.org Objet : Mirror Maker Errors Hello, i running mirror maker for mirroring data from one cluster to another. Now i get this error in log Feb 25 22:38:56 ld4-27 MirrorMaker[54827]: [2018-02-25 22:38:56,914] ERROR Error when sending message to topic rc.exchange.jpy with key: 29 bytes, value: 153 bytes with error: (org.apache.kafka.clients.producer.internals.ErrorLoggingCallback) Feb 25 22:38:56 ld4-27 MirrorMaker[54827]: org.apache.kafka.common.errors.TimeoutException: Expiring 82 record(s) for rc.exchange.jpy-0: 50381 ms has passed since last append I use maxintvalue retries in producer config. Why this message occured. I cant detect any issues. -- Best regards, *Oleg Danilovich*
