[ISSUE #8970] Remove redundant heartbeats

[weihubeats]

#8970

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 25.00000% with 3 lines in your changes missing coverage. Please review.

Project coverage is 47.60%. Comparing base (9202de3) to head (2bed031).
Report is 20 commits behind head on develop.

Files with missing lines	Patch %	Lines
...rocketmq/client/impl/factory/MQClientInstance.java	25.00%	3 Missing ⚠️

Additional details and impacted files

@@              Coverage Diff              @@
##             develop    #8971      +/-   ##
=============================================
- Coverage      47.73%   47.60%   -0.13%     
+ Complexity     11774    11752      -22     
=============================================
  Files           1304     1304              
  Lines          91118    91199      +81     
  Branches       11692    11716      +24     
=============================================
- Hits           43495    43418      -77     
- Misses         42264    42400     +136     
- Partials        5359     5381      +22     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[RongtongJin]

LGTM

[qianye1001]

it is used to make sure any proxy can receive heartbeat request before receive send or pull request when some proxy services share LB

[lizhimins]

Please look the previous comment

[weihubeats]

it is used to make sure any proxy can receive heartbeat request before receive send or pull request when some proxy services share LB

Can you give me an example?

[qianye1001]

it is used to make sure any proxy can receive heartbeat request before receive send or pull request when some proxy services share LB

Can you give me an example?

you can see how GO_AWAY works in remoting protocol to make proxy shutdown gracefully
#8599

when client receive GO_AWAY from proxy, client will reconnect proxy and send heartbeat first.

[weihubeats]

We can track connections to the broker where the initial connection triggers client-initiated heartbeats, while reconnections due to go away will have their heartbeats triggered by the onChannelActive even

[weihubeats]

@qianye1001 How does this sit with you

[RongtongJin]

LGTM

Sorry, I didn't realize it would affect the lossless release code

[qianye1001]

Heartbeats are sent to the broker during client start, rebalancing, scheduled tasks, and Channel creation. A locking is used to prevent duplicate heartbeat transmissions. This is a reasonable solution. Add more complex logic to ensure no additional heartbeats are sent would lead to difficulties in understanding and increased maintenance costs.

if only need prevent warning log, it is more recommended to ignore locking failure or check lock stats before send Heartbeat when Channel creation.

[weihubeats]

Heartbeats are sent to the broker during client start, rebalancing, scheduled tasks, and Channel creation. A locking is used to prevent duplicate heartbeat transmissions. This is a reasonable solution. Add more complex logic to ensure no additional heartbeats are sent would lead to difficulties in understanding and increased maintenance costs.

if only need prevent warning log, it is more recommended to ignore locking failure or check lock stats before send Heartbeat when Channel creation.

I have added a strictLockMode parameter to the sendHeartbeatToBroker method to control whether to log lock acquisition failures. When strictLockMode is true, the system will log lock acquisition failures; when false, these failures will be silently ignored.

[qianye1001]

LGTM

[qianye1001]

@lizhimins

[weihubeats]

@lizhimins

[weihubeats]

@lizhimins Can you help me review this merge request