Fix logger lever from warn to debug when uancked threshold is limited

[wolfstudy]

Signed-off-by: xiaolongran xiaolongran@tencent.com

Motivation

When unacked messages reached to max unacked messages, the following logs will be quickly output to the log file, causing the broker's log file to switch within a short period of time.

Modifications

Fix logger lever from warn to debug when uancked threshold is limited

[merlimat]

I believe in this case the log should really be at debug level. The information is already available in the topic stats for debugging purposes.

[wolfstudy]

I believe in this case the log should really be at debug level. The information is already available in the topic stats for debugging purposes.

Yes, when the unack reaches the maximum threshold, we can get the number of unackedMessages through pulsar-admin topics stats, and get the maxUnackedMessage threshold from pulsar-admin namespace policies. But when this threshold is triggered, Dispatcher read is blocked. This behavior is actively triggered by the broker, which is not transparent enough to the client-side. We still need a more intuitive form to quickly locate this problem. It seems a reasonable way to output the specified log information to the broker log file.

[wolfstudy]

ping @merlimat @codelipenghui @hangc0276 @315157973 PTAL thanks

[wolfstudy]

Which approach do we prefer here:

1. We directly change the log level to debug
2. Introduce log flow control to limit the frequency of log output

[hangc0276]

IMO，we'd better change the log level from warn to debug, and can get the unack info from topic stats.

[wolfstudy]

IMO，we'd better change the log level from warn to debug, and can get the unack info from topic stats.

OK i will fix this change

[wolfstudy]

ping @merlimat @codelipenghui @hangc0276 @315157973 PTAL thanks

[wolfstudy]

/pulsarbot run-failure-checks

[wolfstudy]

/pulsarbot run-failure-checks

[Shawyeok]

I've open a follow up PR #15767 to change other log lines, PTAL.