[Bug] How to solve or optimize this problem [Client memory buffer is full]

[rogueVpedro]

Search before asking

• I searched in the issues and found nothing similar.

Version

2.10.1 master

Minimal reproduce step

When we call the client for a long time, we find that the client is down and cannot send messages.
Then we got an exception while calling the client producer to send messages. (Maybe the frequency of use is too high)

Before asking, we have checked the relevant issues #15724 and #15691，but we still don't know how to avoid this situation.
And this is our process of creating producers

For the configuration of these two parameters(maxPendingMessages and memoryLimit), we set the default values . Btw，The JVM memory of the producer client is 10g and our server memory is 24g.When a problem occurs, out of heap memory is sufficient.
So，how to solve or optimize this problem [Client memory buffer is full]

What did you expect to see?

Avoid this problem, or recover to the original normal situation when the client hangs up.

What did you see instead?

Failing

Anything else?

We sincerely hope to solve this problem,

Are you willing to submit a PR?

• I'm willing to submit a PR!

[Technoboy-]

The default memoryLimit is 64M.
Try to set a larger value.

PulsarClient.builder()
                    .memoryLimit(xxx, SizeUnit.MEGA_BYTES)

[gaozhangmin]

If you encountered send error, like timeout, there were multiple bug about client memory release, please reference
#16837
#16915
#16957
#16958
#16971

[rogueVpedro]

默认内存限制为 64M。 尝试设置更大的值。

PulsarClient.builder()
                    .memoryLimit(xxx, SizeUnit.MEGA_BYTES)

In fact, the size of each message is controlled within 64M. We judge the size of each message before sending it to the producer queue. But still found this problem again

[rogueVpedro]

If you encountered send error, like timeout, there were multiple bug about client memory release, please reference #16837 #16915 #16957 #16958 #16971

OK, I will refer to the solutions to these bugs, but I am not sure whether they are useful.

[github-actions]

The issue had no activity for 30 days, mark with Stale label.

[Jason918]

In fact, the size of each message is controlled within 64M. We judge the size of each message before sending it to the producer queue. But still found this problem again

@rogueVpedro FYI: 64MB is the total memory limit, not for each message.

[github-actions]

The issue had no activity for 30 days, mark with Stale label.