[Bug 66202] New: Need different group of worker thread to process the request

[bu...@apache.org]

https://bz.apache.org/bugzilla/show_bug.cgi?id=66202

            Bug ID: 66202
           Summary: Need different group of worker thread to process the
                    request
           Product: Tomcat 9
           Version: unspecified
          Hardware: PC
                OS: All
            Status: NEW
          Severity: major
          Priority: P2
         Component: Connectors
          Assignee: dev@tomcat.apache.org
          Reporter: iihero@qq.com
  Target Milestone: -----

Perhaps this can be considered as a Feature instead of a bug.
The scenario is in the Cloud Foundry platform. 

The healthcheck endpoint "/manage/info" is the same with the other all user
business endpoint even if they are registered with different priority.

When the business endpoint has a long time response and the server is in a high
concurrency,  the the healthcheck endpoint will lose the chance to respond in
time because all the worker thread is busy on long processing of the other
endpoint. 
Finally CF container will crash with OOM (137).

One important reason is that all the endponts are processed in the same group
of worker thread.

My thinking is that,  whether tomcat can provide at least 2 groups of worker
thread to work on different group of endpoints. 
thus the internal group of worker thread which processes healthcheck will never
be blocked and this can ensure the server is alive. 

BR,
Sean

-- 
You are receiving this mail because:
You are the assignee for the bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org

[Bug 66202] Need different group of worker thread to process the request

[bu...@apache.org]

https://bz.apache.org/bugzilla/show_bug.cgi?id=66202

Mark Thomas <ma...@apache.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |INVALID
             Status|NEW                         |RESOLVED

--- Comment #1 from Mark Thomas <ma...@apache.org> ---
If an application is blocking all processing threads and new requests can't be
processed the health check should fail.

OOME is a strong indicator of either an application memory leak or an
undersized JVM.

If you really want do this (and I'd strongly recommend that you don't) then you
can use a dedicated Connector and perform the health check via a dedicated
port.

-- 
You are receiving this mail because:
You are the assignee for the bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org

[Bug 66202] Need different group of worker thread to process the request

[bu...@apache.org]

https://bz.apache.org/bugzilla/show_bug.cgi?id=66202

Sean <ii...@qq.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |iihero@qq.com

-- 
You are receiving this mail because:
You are the assignee for the bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org