Memory leak detected in BtyeBuf.release()

[dlvenable]

We use Armeria and recently saw some error logs like the following:

2023-12-27T23:16:48.364 [armeria-common-worker-epoll-3-2] ERROR       io.netty.util.ResourceLeakDetector - LEAK: ByteBuf.release() was not called before it's garbage-collected. See https://netty.io/wiki/reference-counted-objects.html for more information.

Is this a bug in Armeria? Could this be caused by a programming error in our code? And if so, how do we best debug this type of situation?

We are currently using Armeria 1.25.2.

Here is our internal issue: opensearch-project/data-prepper#3912

[ikhoon]

A known leak has been fixed in #5227 with Armeria 1.26.0
It is difficult to know the cause of the leak with the given information. Could you try to raise the leak detection level to find the leak locations?

[dlvenable]

@ikhoon , Thank you for the response. We will update the Armeria version to see if that helps. And we'll raise the leak detection level to get some more information.

At a high-level, can our code be responsible for this? I don't see our code needing to use the ByteBuf, but can we cause this by handling the request incorrectly?

[ikhoon]

If you didn't use ByteBuf directly by subscribing to HttpRequest with WITH_POOLED_OBJECTS, it would be a bug in the Armeria internal.

armeria/core/src/main/java/com/linecorp/armeria/common/stream/SubscriptionOption.java

Lines 43 to 49 in 74eb55f

	* (Advanced users only) To receive the pooled {@link HttpData} as is, without making a copy.
	* If you don't know what this means, do not specify this when you subscribe the {@link StreamMessage}.
	*
	* @see PooledObjects
	*/
	@UnstableApi
	WITH_POOLED_OBJECTS