enablePeriodicRefresh has problematic behavior when service is shut down(close)

[Kvicii]

Bug Report

Current Behavior

For example: I have 2 classes, A and B, and the close method is as follows.
A will cycle through multiple RedisSenders (B) and finally release ClientResources. At the same time, ClusterTopologyRefreshOptions is set as follows:

.enablePeriodicRefresh(10 seconds)
.enableAllAdaptiveRefreshTriggers()
.closeStaleConnections(true)
.dynamicRefreshSources(true)

I found that it might be because the netty thread was not closed in time, and the following error occurred:

// A
public void close() {
    for (RedisSender sender : RedisSender List) {
        sender.close();
    }

    if (clientResources != null) {
        clientResources.shutdown();
    }
}

// B RedisSender
public void close() {
    StatefulRedisClusterConnection<String, byte[]>.close();
    RedisCluster.close();
}

Expected behavior/code

After the ClientResources are closed, the tasks being processed should be discarded

Environment

• Lettuce version(s): 6.3.2-RELEASE

[tishun]

This seems to be the same issue as the one described in #2904

[Kvicii]

@tishun yes, but I'm not sure if it can be fixed soon, as #2985 mentioned, I'm not sure if it works.

[tishun]

Closing in favor of #2904
Let's continue the discussion there.

If this is another problem, please reopen and describe how it is different so we can try and address it too.

[tishun]

@tishun yes, but I'm not sure if it can be fixed soon, as #2985 mentioned, I'm not sure if it works.

Not fixed yet.

How big of a problem is it for you? Because it seems to me like it should not have a big effect on the application, besides the log messages?

[tishun]

@tishun yes, but I'm not sure if it can be fixed soon, as #2985 mentioned, I'm not sure if it works.

Not fixed yet.

How big of a problem is it for you? Because it seems to me like it should not have a big effect on the application, besides the log messages?

To elaborate more on my question:

• if there is a functional regression caused by this problem then we would increase it's priority
• if not I would have to postpone it for now, as there are actual functional bugs that are with higher priority

As always if you believe you can contribute a solution please send out a PR with your idea.

[Kvicii]

@tishun No significant impact has been found so far, but for a stable service, some error logs should be avoided as much as possible.