[jetty-dev] [jira] (JETTY-1527) QueuedThreadPool Anomaly

David Wade
commented on JETTY-1527

QueuedThreadPool Anomaly

Thanks Jan,

We actually noticed this problem in our running application on Jetty (with commercial support from Intalio). However, this was not when used with the HTTP connector but when reusing for our own thread pools within our applications (so not a support issue). Our applications are running on multiple reasonably big servers (24 actual cores per server). Within our applications, we use the pool patched.

A bit more on the current implementation, those two lines are currently before the potentially blocking call to _jobs.offer(job). This means on heavily used and threaded applications, in computing terms there can be a large amount of time between the snapshot of those values and making the decision as to whether to grow the pool or not. If the snapshot is performed after the call there is a very compute small time inbetween (baring timeslicing). But more importantly, there is practically no chance the pool will have failed to grow when it needed too.

I understand the reluctance to fix/change, but suspect there may be more than subtle effects when changing to disruptors. In the meantime, this leaves a potential erratic latancy issue for Jetty users unresolved, perhaps for years before they can upgrade to 9.1...

This message is automatically generated by JIRA.

If you think it was sent incorrectly, please contact your JIRA administrators .

For more information on JIRA, see: http://www.atlassian.com/software/jira

---------------------------------------------------------------------
To unsubscribe from this list, please visit:

http://xircles.codehaus.org/manage_email