[jira] [Commented] (KAFKA-1188) Stale LeaderAndIsr request could be handled by the broker on Controller failover
[ https://issues.apache.org/jira/browse/KAFKA-1188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13898648#comment-13898648 ] Guozhang Wang commented on KAFKA-1188: -- Updated reviewboard against branch origin/trunk > Stale LeaderAndIsr request could be handled by the broker on Controller > failover > > > Key: KAFKA-1188 > URL: https://issues.apache.org/jira/browse/KAFKA-1188 > Project: Kafka > Issue Type: Bug >Reporter: Guozhang Wang > Attachments: KAFKA-1188.patch, KAFKA-1188_2013-12-24_11:07:12.patch, > KAFKA-1188_2014-01-02_10:43:54.patch, KAFKA-1188_2014-01-15_16:43:30.patch, > KAFKA-1188_2014-01-16_11:06:56.patch, KAFKA-1188_2014-01-16_15:02:22.patch, > KAFKA-1188_2014-01-21_14:01:22.patch, KAFKA-1188_2014-02-11_17:48:06.patch, > KAFKA-1188_2014-02-11_17:50:06.patch > > > ... which will cause the broker to truncate log as responding in > makeFollower, and hence lose data. > One procedure to produce this issue: > 3 brokers, 3 partitions, replication factor = 3. > 1. Broker 1 is the original controller. > 2. Broker 3 lost registration, 1 send LeaderAndIsr to 1 for isr shrinking > (leader of partition 1), but not to 2 since it is not the leader. > 3. Broker 1 dies, new controller 2 sends all LeaderAndIsr to 2 and 3 > (resumed), and 2 handles the makeFollower, truncates data. -- This message was sent by Atlassian JIRA (v6.1.5#6160)
[jira] [Commented] (KAFKA-1188) Stale LeaderAndIsr request could be handled by the broker on Controller failover
[ https://issues.apache.org/jira/browse/KAFKA-1188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13898646#comment-13898646 ] Guozhang Wang commented on KAFKA-1188: -- Updated reviewboard against branch origin/trunk > Stale LeaderAndIsr request could be handled by the broker on Controller > failover > > > Key: KAFKA-1188 > URL: https://issues.apache.org/jira/browse/KAFKA-1188 > Project: Kafka > Issue Type: Bug >Reporter: Guozhang Wang > Attachments: KAFKA-1188.patch, KAFKA-1188_2013-12-24_11:07:12.patch, > KAFKA-1188_2014-01-02_10:43:54.patch, KAFKA-1188_2014-01-15_16:43:30.patch, > KAFKA-1188_2014-01-16_11:06:56.patch, KAFKA-1188_2014-01-16_15:02:22.patch, > KAFKA-1188_2014-01-21_14:01:22.patch, KAFKA-1188_2014-02-11_17:48:06.patch > > > ... which will cause the broker to truncate log as responding in > makeFollower, and hence lose data. > One procedure to produce this issue: > 3 brokers, 3 partitions, replication factor = 3. > 1. Broker 1 is the original controller. > 2. Broker 3 lost registration, 1 send LeaderAndIsr to 1 for isr shrinking > (leader of partition 1), but not to 2 since it is not the leader. > 3. Broker 1 dies, new controller 2 sends all LeaderAndIsr to 2 and 3 > (resumed), and 2 handles the makeFollower, truncates data. -- This message was sent by Atlassian JIRA (v6.1.5#6160)
[jira] [Commented] (KAFKA-1188) Stale LeaderAndIsr request could be handled by the broker on Controller failover
[ https://issues.apache.org/jira/browse/KAFKA-1188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13877906#comment-13877906 ] Guozhang Wang commented on KAFKA-1188: -- Updated reviewboard https://reviews.apache.org/r/16360/ against branch origin/trunk > Stale LeaderAndIsr request could be handled by the broker on Controller > failover > > > Key: KAFKA-1188 > URL: https://issues.apache.org/jira/browse/KAFKA-1188 > Project: Kafka > Issue Type: Bug >Reporter: Guozhang Wang > Attachments: KAFKA-1188.patch, KAFKA-1188_2013-12-24_11:07:12.patch, > KAFKA-1188_2014-01-02_10:43:54.patch, KAFKA-1188_2014-01-15_16:43:30.patch, > KAFKA-1188_2014-01-16_11:06:56.patch, KAFKA-1188_2014-01-16_15:02:22.patch, > KAFKA-1188_2014-01-21_14:01:22.patch > > > ... which will cause the broker to truncate log as responding in > makeFollower, and hence lose data. > One procedure to produce this issue: > 3 brokers, 3 partitions, replication factor = 3. > 1. Broker 1 is the original controller. > 2. Broker 3 lost registration, 1 send LeaderAndIsr to 1 for isr shrinking > (leader of partition 1), but not to 2 since it is not the leader. > 3. Broker 1 dies, new controller 2 sends all LeaderAndIsr to 2 and 3 > (resumed), and 2 handles the makeFollower, truncates data. -- This message was sent by Atlassian JIRA (v6.1.5#6160)
[jira] [Commented] (KAFKA-1188) Stale LeaderAndIsr request could be handled by the broker on Controller failover
[ https://issues.apache.org/jira/browse/KAFKA-1188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13874112#comment-13874112 ] Guozhang Wang commented on KAFKA-1188: -- Updated reviewboard https://reviews.apache.org/r/16360/ against branch origin/trunk > Stale LeaderAndIsr request could be handled by the broker on Controller > failover > > > Key: KAFKA-1188 > URL: https://issues.apache.org/jira/browse/KAFKA-1188 > Project: Kafka > Issue Type: Bug >Reporter: Guozhang Wang > Attachments: KAFKA-1188.patch, KAFKA-1188_2013-12-24_11:07:12.patch, > KAFKA-1188_2014-01-02_10:43:54.patch, KAFKA-1188_2014-01-15_16:43:30.patch, > KAFKA-1188_2014-01-16_11:06:56.patch, KAFKA-1188_2014-01-16_15:02:22.patch > > > ... which will cause the broker to truncate log as responding in > makeFollower, and hence lose data. > One procedure to produce this issue: > 3 brokers, 3 partitions, replication factor = 3. > 1. Broker 1 is the original controller. > 2. Broker 3 lost registration, 1 send LeaderAndIsr to 1 for isr shrinking > (leader of partition 1), but not to 2 since it is not the leader. > 3. Broker 1 dies, new controller 2 sends all LeaderAndIsr to 2 and 3 > (resumed), and 2 handles the makeFollower, truncates data. -- This message was sent by Atlassian JIRA (v6.1.5#6160)
[jira] [Commented] (KAFKA-1188) Stale LeaderAndIsr request could be handled by the broker on Controller failover
[ https://issues.apache.org/jira/browse/KAFKA-1188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13873785#comment-13873785 ] Guozhang Wang commented on KAFKA-1188: -- Updated reviewboard https://reviews.apache.org/r/16360/ against branch origin/trunk > Stale LeaderAndIsr request could be handled by the broker on Controller > failover > > > Key: KAFKA-1188 > URL: https://issues.apache.org/jira/browse/KAFKA-1188 > Project: Kafka > Issue Type: Bug >Reporter: Guozhang Wang > Attachments: KAFKA-1188.patch, KAFKA-1188_2013-12-24_11:07:12.patch, > KAFKA-1188_2014-01-02_10:43:54.patch, KAFKA-1188_2014-01-15_16:43:30.patch, > KAFKA-1188_2014-01-16_11:06:56.patch > > > ... which will cause the broker to truncate log as responding in > makeFollower, and hence lose data. > One procedure to produce this issue: > 3 brokers, 3 partitions, replication factor = 3. > 1. Broker 1 is the original controller. > 2. Broker 3 lost registration, 1 send LeaderAndIsr to 1 for isr shrinking > (leader of partition 1), but not to 2 since it is not the leader. > 3. Broker 1 dies, new controller 2 sends all LeaderAndIsr to 2 and 3 > (resumed), and 2 handles the makeFollower, truncates data. -- This message was sent by Atlassian JIRA (v6.1.5#6160)
[jira] [Commented] (KAFKA-1188) Stale LeaderAndIsr request could be handled by the broker on Controller failover
[ https://issues.apache.org/jira/browse/KAFKA-1188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13872882#comment-13872882 ] Guozhang Wang commented on KAFKA-1188: -- Updated reviewboard https://reviews.apache.org/r/16360/ against branch origin/trunk > Stale LeaderAndIsr request could be handled by the broker on Controller > failover > > > Key: KAFKA-1188 > URL: https://issues.apache.org/jira/browse/KAFKA-1188 > Project: Kafka > Issue Type: Bug >Reporter: Guozhang Wang > Attachments: KAFKA-1188.patch, KAFKA-1188_2013-12-24_11:07:12.patch, > KAFKA-1188_2014-01-02_10:43:54.patch, KAFKA-1188_2014-01-15_16:43:30.patch > > > ... which will cause the broker to truncate log as responding in > makeFollower, and hence lose data. > One procedure to produce this issue: > 3 brokers, 3 partitions, replication factor = 3. > 1. Broker 1 is the original controller. > 2. Broker 3 lost registration, 1 send LeaderAndIsr to 1 for isr shrinking > (leader of partition 1), but not to 2 since it is not the leader. > 3. Broker 1 dies, new controller 2 sends all LeaderAndIsr to 2 and 3 > (resumed), and 2 handles the makeFollower, truncates data. -- This message was sent by Atlassian JIRA (v6.1.5#6160)
[jira] [Commented] (KAFKA-1188) Stale LeaderAndIsr request could be handled by the broker on Controller failover
[ https://issues.apache.org/jira/browse/KAFKA-1188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13860586#comment-13860586 ] Guozhang Wang commented on KAFKA-1188: -- Updated reviewboard https://reviews.apache.org/r/16360/ against branch origin/trunk > Stale LeaderAndIsr request could be handled by the broker on Controller > failover > > > Key: KAFKA-1188 > URL: https://issues.apache.org/jira/browse/KAFKA-1188 > Project: Kafka > Issue Type: Bug >Reporter: Guozhang Wang > Attachments: KAFKA-1188.patch, KAFKA-1188_2013-12-24_11:07:12.patch, > KAFKA-1188_2014-01-02_10:43:54.patch > > > ... which will cause the broker to truncate log as responding in > makeFollower, and hence lose data. > One procedure to produce this issue: > 3 brokers, 3 partitions, replication factor = 3. > 1. Broker 1 is the original controller. > 2. Broker 3 lost registration, 1 send LeaderAndIsr to 1 for isr shrinking > (leader of partition 1), but not to 2 since it is not the leader. > 3. Broker 1 dies, new controller 2 sends all LeaderAndIsr to 2 and 3 > (resumed), and 2 handles the makeFollower, truncates data. -- This message was sent by Atlassian JIRA (v6.1.5#6160)
[jira] [Commented] (KAFKA-1188) Stale LeaderAndIsr request could be handled by the broker on Controller failover
[ https://issues.apache.org/jira/browse/KAFKA-1188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13856426#comment-13856426 ] Guozhang Wang commented on KAFKA-1188: -- Updated reviewboard https://reviews.apache.org/r/16360/ against branch origin/trunk > Stale LeaderAndIsr request could be handled by the broker on Controller > failover > > > Key: KAFKA-1188 > URL: https://issues.apache.org/jira/browse/KAFKA-1188 > Project: Kafka > Issue Type: Bug >Reporter: Guozhang Wang > Attachments: KAFKA-1188.patch, KAFKA-1188_2013-12-24_11:07:12.patch > > > ... which will cause the broker to truncate log as responding in > makeFollower, and hence lose data. > One procedure to produce this issue: > 3 brokers, 3 partitions, replication factor = 3. > 1. Broker 1 is the original controller. > 2. Broker 3 lost registration, 1 send LeaderAndIsr to 1 for isr shrinking > (leader of partition 1), but not to 2 since it is not the leader. > 3. Broker 1 dies, new controller 2 sends all LeaderAndIsr to 2 and 3 > (resumed), and 2 handles the makeFollower, truncates data. -- This message was sent by Atlassian JIRA (v6.1.5#6160)
[jira] [Commented] (KAFKA-1188) Stale LeaderAndIsr request could be handled by the broker on Controller failover
[ https://issues.apache.org/jira/browse/KAFKA-1188?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13852541#comment-13852541 ] Guozhang Wang commented on KAFKA-1188: -- Created reviewboard https://reviews.apache.org/r/16360/ against branch origin/trunk > Stale LeaderAndIsr request could be handled by the broker on Controller > failover > > > Key: KAFKA-1188 > URL: https://issues.apache.org/jira/browse/KAFKA-1188 > Project: Kafka > Issue Type: Bug >Reporter: Guozhang Wang > Attachments: KAFKA-1188.patch > > > ... which will cause the broker to truncate log as responding in > makeFollower, and hence lose data. > One procedure to produce this issue: > 3 brokers, 3 partitions, replication factor = 3. > 1. Broker 1 is the original controller. > 2. Broker 3 lost registration, 1 send LeaderAndIsr to 1 for isr shrinking > (leader of partition 1), but not to 2 since it is not the leader. > 3. Broker 1 dies, new controller 2 sends all LeaderAndIsr to 2 and 3 > (resumed), and 2 handles the makeFollower, truncates data. -- This message was sent by Atlassian JIRA (v6.1.4#6159)