* Make configPath unable to update at runtime
* Make configPath unable to update at runtime
* Add ASF header
* Add ASF header
* Pass the check style
* Polish the response code
* Polish the response code
* 1. When compute the confirmOffset, dismiss the ackOffset of new connections. 2. When compute the confirmOffset, use getConfirmOffsetDirectly() to avoid the endless calling.
* use the calculated slaveAckOffset
* optimize the logic.
* bugfix broker boot succes but get fail ip addr
bug:
broker ip addr(IPV4) get fail after broker start up;
fix:
add compare ,continue when ip is start with '0.' ;
like :
The broker[broker-a, 0.0.1.1:10911] boot success. serializeType=JSON and name server is 127.0.0.1:9876;
* add timerstore nil condition when timer no enable
add timerstore nil condition when timer no enable
* close timer schedule thread when timer no enable
close timer schedule thread when timer no enable
* format code
format code
* Refector PopLongPollingService
* Use enum PollingResult to replace int value
* Move PopLongPollingService to upper level
* Move wakeUp and polling method into PopLongPollingService
* NotificationProcessor use PopLongPollingService
* Add BORN_TIME check
* add NotificationIT
* 1. Fix the bug that confirmOffset may be larger than ReputOffset in some cases. 2. Optimize the logic when truncate dirty files.
* Add more logs while expanding SyncStateSet.
* reset the confirmOffset to the truncated offset.
* optimize the log.
* move setConfirmOffset() process into commitLog.
* add the judge.
* When compute confirmOffset, judge whether the slaves in syncStateSet all connect to the master.
* fix the brokerId to brokerControllerId.
* optimize the code: 1. Set brokerId when the replicasManager inited. 2. Rename the confusing name.
* set haService's brokerControllerId after registered.
* set the haService's brokerControllerId when the ReplicasManager start without brokerIdentity file.
* fix(tieredstorage): fix the bug that not all message fetch request is processed via tieredStorage when enable FORCE
1. fix the bug that not all message fetch request is processed via
tieredStorage when enable FORCE
* style(tieredstorage): use rmq code style
1. use rmq code style
* feat(tieredstorage): add more comments for `TieredStorageLevel`
1. add more comments for `TieredStorageLevel`
* Fix the issue that expectLogicOffset is greater than currentLogicOffset in consumeQueue build when the message is illegal
* Add new UT
* Fix bug that UT can not pass
* Polish the variable name
* Polish the comment
* Add more comments
* [ISSUE #6537] Fix bug MessageStoreConfig.haListenPort item config not effective
* rerun
* [ISSUE #6537] fix bug, if user not config haListenPort, then safe set haListenPort value as listenPort +1
* rerun
* (fix): logback config path seperator and brokerLogDir default value missing
* (fix): replace / with file.seperator in proxy test logback config
---------
Co-authored-by: Robin Han <robin@Robindendiannao.lan>
* [#6525] Make ConsumeQueueInterface extends from FileQueueLifeCycle
* [apache#6525] forced typecast is removed
* [#6525] remove all the forced convert
* Add a description of the Apache RocketMQ E2E repository in README.md
* Update README.md
Co-authored-by: Aaron Ai <yangkun.ayk@gmail.com>
---------
Co-authored-by: Aaron Ai <yangkun.ayk@gmail.com>
rongtong