[mod_cluster-issues] [JBoss JIRA] (MODCLUSTER-383) Session draining broken: requests counting broken on load-balancer

[Radoslav Husar (JIRA)]

    [ https://issues.jboss.org/browse/MODCLUSTER-383?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12941306#comment-12941306 ] 

Radoslav Husar commented on MODCLUSTER-383:
-------------------------------------------

I have verified the fix, it works for contexts.

I also noticed that other 2 values are wrong for the node:

{noformat}Balancer: mycluster,LBGroup: ,Flushpackets: Off,Flushwait: 10000,Ping: 10000000,Smax: 2,Ttl: 60000000,Status: OK,Elected: 297148,Read: 0,Transferred: 0,Connected: 5,Load: 74{noformat}

Namely: elected is at 297,148 even though the test sent 300,000 requets precisely. Connected is at 5, even though the server is down and all clients are off. Should we be worried about these too?
                
> Session draining broken: requests counting broken on load-balancer
> ------------------------------------------------------------------
>
>                 Key: MODCLUSTER-383
>                 URL: https://issues.jboss.org/browse/MODCLUSTER-383
>             Project: mod_cluster
>          Issue Type: Bug
>    Affects Versions: 1.3.0.Alpha1
>         Environment: Fedora 20, 64 bit, httpd 2.4.6 + mod_cluster master (21ceed3c219fc3ad743b361cafd1097ebac19dfe)
>            Reporter: Radoslav Husar
>            Assignee: Jean-Frederic Clere
>            Priority: Critical
>             Fix For: 1.3.0.Alpha2
>
>
> The request counting is broken. Looks like synchronization problem with dirty cached reads.
> Steps to reproduce:
> # start AS with some context
> # start LB
> # start 2 or more load driver threads
> # number of requests on that context goes to higher values than 2, eventually and slowly increasing
> On the AS it manifests as:
> {noformat}
> 19:44:14,160 WARN  [org.jboss.modcluster] (MSC service thread 1-7) MODCLUSTER000022: Failed to drain 57 remaining pending requests from default-host:/clusterbench within 10.0 seconds
> {noformat}

--
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