[ https://issues.apache.org/jira/browse/KAFKA-14739?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17692452#comment-17692452 ]
Luke Chen commented on KAFKA-14739: ----------------------------------- [~colinshaw] , Have you tried to run with the latest version of kafka client? We did some improvement to it before. Please check if this issue still existed. Thanks. > Kafka consumer reading messages out of order after a rebalance > -------------------------------------------------------------- > > Key: KAFKA-14739 > URL: https://issues.apache.org/jira/browse/KAFKA-14739 > Project: Kafka > Issue Type: Bug > Affects Versions: 2.6.0 > Reporter: Colin Shaw > Priority: Major > Attachments: image-2023-02-21-16-45-10-464.png > > > We are seeing often when our consumers are running and a rebalance is > triggered, some partitions do not resume at the correct offset causing a > large number of events to be skipped. > Our understanding is that within a partition events should also be read in > chronological order. > > These consumers run once a day and we see that when they resume again the > next day it does return to the missed events. > > An example we investigated: > {quote}We see for partition 88 starting processing around Offset 14001. > Events were processed until Offset 14059 but we could see Offset 14060 was > not processed from out logs. > Upon investigation we could see a rebalance had occured after offset 14059 > and afterwards the next event processed was offset 14190 causing about 130 > events to be skipped on this partition. > The subsequent day we it did resume at Offset 14060 > {quote} > > -- This message was sent by Atlassian Jira (v8.20.10#820010)