[ https://issues.jboss.org/browse/WFCORE-2620?page=com.atlassian.jira.plugi... ] Romain Pelisse edited comment on WFCORE-2620 at 5/8/17 6:29 AM: ---------------------------------------------------------------- [~ctomc] following your comment (and Brian's) on my [PR|https://github.com/wildfly/wildfly-core/pull/2330] for [WFCORE-2620|https://issues.jboss.org/browse/WFCORE-2620], I've looked at how you achieved a proper runtime computed value of the workers. AFAIU, you simply used the [MBean|https://github.com/wildfly/wildfly-core/blob/master/io/subsystem/sr...], provided by Xnio(1) to get the proper value. However, afaiu, there is no value related to buffer in this MBean. There is an other MBean, java.nio, providing info in the BufferPool(2), however the attributes available there (TotalCapacity and Count) does not provides us with the metrics we need here (buffer-size and buffers-per-slice). AFAIU, the metrics associated to those are computer in the [BufferPoolResourceDefinition class|https://github.com/wildfly/wildfly-core/blob/master/io/subsystem/sr...], and maybe override using the subsystem configuration in [the standalone.xm|https://github.com/wildfly/wildfly-core/blob/master/io/subs...]. My naive approach here - that I want to double check with you, is : Should I create a MBean for the buffer attribute - to mimic the approach of the worker ? If so I assume we'll move the [buffers "default computation"|https://github.com/wildfly/wildfly-core/blob/master/io/...] into the MBean (right?). Or am I misunderstanding or missing something here ? Let me know what you think about this (and if my analysis is correct) MX object names: * (1) ObjectName: org.xnio:tyep=Xnio.provider="nio,work"="default" * (2) ObjectName: java.nio:type=BufferPool,name=direct] was (Author: rpelisse): [~ctomc] following your comment (and Brian's) on my [PR|https://github.com/wildfly/wildfly-core/pull/2330] for [WFCORE-2620|https://issues.jboss.org/browse/WFCORE-2620], I've looked at you achieved a proper runtime computed value of the workers. AFAIU, you simply used the [MBean|https://github.com/wildfly/wildfly-core/blob/master/io/subsystem/sr...], provided by Xnio(1) to get the proper value. However, there is no value related to buffer in this MBean. There is an other MBean, java.nio, providing info in the BufferPool(2), however the attributes available there (TotalCapacity and Count) does not provides us with the metrics we need here (buffer-size and buffers-per-slice). AFAIU, the metrics associated to those are computer in the [BufferPoolResourceDefinition class|https://github.com/wildfly/wildfly-core/blob/master/io/subsystem/sr...], and maybe override using the subsystem configuration in [the standalone.xm|https://github.com/wildfly/wildfly-core/blob/master/io/subs...]. My naive approach here - that I want to double check with you, is : Should I create a MBean for the buffer attribute - to mimic the approach of the worker ? If so I assume we'll move the [buffers "default computation"|https://github.com/wildfly/wildfly-core/blob/master/io/...] into the MBean (right?). Or am I misunderstanding or missing something here ? Let me know what you think about this (and if my analysis is correct) MX object names: * (1) ObjectName: org.xnio:tyep=Xnio.provider="nio,work"="default" * (2) ObjectName: java.nio:type=BufferPool,name=direct] -- This message was sent by Atlassian JIRA (v7.2.3#72005)