[
https://issues.apache.org/jira/browse/DIRMINA-618?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Christopher Hurst updated DIRMINA-618:
--------------------------------------
Description:
The latest version of Mina (I believe 1.1.7 worked see link) has issues with
Glassfish when deployed as inbound JCA as reported else where (Has this been
worked on ? / explained ?) ...
http://www.nabble.com/MINA-and-Glassfish-td18767392.html
The non working machine shows
java.nio.channels.spi.SelectorProvider=com.sun.enterprise.ss.provider.ASSelector
(from glassfish)
the working machine shows this as unset, I've tried unsetting this myself but
it seems not that simple, I guess its a case of when it has to be unset to take
effect.
On the failing machine attempting to connect via a NioSocketConnector
establishes a connection but it then goes horribly wrong usually with a null
pointer exception in ...
org.apache.mina.core.polling.AbstractPollingIoConnector
ConnectionRequest connectionRequest(H handle); returns null which causes the
problem, looks like lower down that the wrong set of keys is being obtained ,
decorated vs undecorated ie right set of keys but they're wrapped by a
decorator the key comparison can't cope with ..
I noticed whilst debugging that interacting with the debugger (breakpoints)
caused the socket to work on my quad core vista box, then by recoding MINA to
ignore the null pointers it all sprang into life and appears stable once
connected, however my dual core XP box still doesn't work, obviously the vista
box is way higher spec which may or may not effect the issue.
I'm trying to bottom this out myself but any feedback would be useful,
particularly if some one has identified the issue already.
I have a none JCA version of my code and it works fine everywhere as is.
I notice using the debugger that the MINA code flow jumps into a glassfish lib
appserv-rt something to do with the selectors but haven't got to the bottom of
this yet.
was:
The latest version of Mina (I believe 1.1.7 worked see link) has issues with
Glassfish when deployed as inbound JCA as reported else where (Has this been
worked on ? / explained ?) ...
http://www.nabble.com/MINA-and-Glassfish-td18767392.html
Attempting to connect via a NioSocketConnector establishes a connection but it
then goes horribly wrong usually with a null pointer exception in ...
org.apache.mina.core.polling.AbstractPollingIoConnector
ConnectionRequest connectionRequest(H handle); returns null which causes the
problem.
I noticed whilst debugging that interacting with the debugger (breakpoints)
caused the socket to work on my quad core vista box, then by recoding MINA to
ignore the null pointers it all sprang into life and appears stable once
connected, however my dual core XP box still doesn't work, obviously the vista
box is way higher spec which may or may not effect the issue.
I'm trying to bottom this out myself but any feedback would be useful,
particaularly if some one has identified the issue already.
I have a none JCA version of my code and it works fine everywhere as is.
I notice using the debugger that the MINA code flow jumps into a glassfish lib
appserv-rt something to do with the selectors but haven't got to the bottom of
this yet.
My two current theorys are a race condition and /or the glassfish jar causing
the issue.
I'm updating this report with more info as I get to the root cause .
> Latest MINA won't work with glassfish as JCA inbound connector
> --------------------------------------------------------------
>
> Key: DIRMINA-618
> URL: https://issues.apache.org/jira/browse/DIRMINA-618
> Project: MINA
> Issue Type: Bug
> Components: Core
> Affects Versions: 2.0.0-M3
> Environment: Windows XP / Vista
> Reporter: Christopher Hurst
> Priority: Blocker
> Original Estimate: 168h
> Remaining Estimate: 168h
>
> The latest version of Mina (I believe 1.1.7 worked see link) has issues with
> Glassfish when deployed as inbound JCA as reported else where (Has this been
> worked on ? / explained ?) ...
> http://www.nabble.com/MINA-and-Glassfish-td18767392.html
> The non working machine shows
> java.nio.channels.spi.SelectorProvider=com.sun.enterprise.ss.provider.ASSelector
> (from glassfish)
> the working machine shows this as unset, I've tried unsetting this myself
> but it seems not that simple, I guess its a case of when it has to be unset
> to take effect.
> On the failing machine attempting to connect via a NioSocketConnector
> establishes a connection but it then goes horribly wrong usually with a null
> pointer exception in ...
> org.apache.mina.core.polling.AbstractPollingIoConnector
> ConnectionRequest connectionRequest(H handle); returns null which causes the
> problem, looks like lower down that the wrong set of keys is being obtained ,
> decorated vs undecorated ie right set of keys but they're wrapped by a
> decorator the key comparison can't cope with ..
> I noticed whilst debugging that interacting with the debugger (breakpoints)
> caused the socket to work on my quad core vista box, then by recoding MINA to
> ignore the null pointers it all sprang into life and appears stable once
> connected, however my dual core XP box still doesn't work, obviously the
> vista box is way higher spec which may or may not effect the issue.
> I'm trying to bottom this out myself but any feedback would be useful,
> particularly if some one has identified the issue already.
> I have a none JCA version of my code and it works fine everywhere as is.
> I notice using the debugger that the MINA code flow jumps into a glassfish
> lib appserv-rt something to do with the selectors but haven't got to the
> bottom of this yet.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
