HDFS-16771. Follow-up for HDFS-16659: JN should tersely print logs about NewerTxnIdException#4882
Conversation
|
💔 -1 overall
This message was automatically generated. |
xkrogen
left a comment
There was a problem hiding this comment.
As a note to others -- see more context on why this change is needed in @ZanderXu's comment here.
This implementation, where we special-case the sinceTxId == highestTxId + 1 situation, doesn't 100% solve the original issue that #4560 was tackling since it can still cause issues if a fast-responding journal is lagging by exactly 1 tx id (as described in this comment).
But I think with the issues you described in your last comment on the PR, I agree that it's not worth it introduce the extra complexity for this extreme corner case. So I am okay with this PR besides the minor comments I left.
...t/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/qjournal/client/QuorumJournalManager.java
Show resolved
Hide resolved
...t/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/qjournal/server/JournalNodeRpcServer.java
Show resolved
Hide resolved
...p-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/qjournal/server/Journal.java
Outdated
Show resolved
Hide resolved
|
🎊 +1 overall
This message was automatically generated. |
|
🎊 +1 overall
This message was automatically generated. |
|
Merged to trunk. Thanks @ZanderXu ! |
…out NewerTxnIdException (apache#4882) Signed-off-by: Erik Krogen <xkrogen@apache.org> Co-authored-by: zengqiang.xu <zengqiang.xu@shopee.com>
JournalNode should tersely print some logs about NewerTxnIdException.