Hi Rich,

Thanks for your response.  I very appreciate it.

>> see if you can specify that no controls are to be sent with the Abandon request
I looked at the JNDI API documentation and I don't see there's a way I can do this.  I did verify by reviewing the OpenJDK source code that it is setting the control as CRITICAL, and it is set in the com.sun.jndi.ldap.NamingEventNotifier class.

>> if you wait for all of the operations and results to be read by your app, JNDI might notice this and just do an Unbind instead of an Abandon.
The test application that I use done things in a sequential order so I believe all the operations and results had already been read.

I did more testings today using the same test I attached in my original e-mail and found out that this problem happen very often but sometime the connection does cleaned up correctly.  Here's two snippets of the errors log with "Connection management" log level enabled.  The difference is that when it does work, I see the "sasl_io_cleanup for connection" debug statement and it appears that slapd tries to release the connection and operation in all cases except often time it doesn't work (when persistent search is used) and the connection get into a CLOSE_WAIT state.

I will do more testing and poke more on the JAVA client.

Thanks,

David

###### WORKING LOG ######
[19/Feb/2010:13:12:49 -0500] - activity on 68r
[19/Feb/2010:13:12:49 -0500] - read activity on 68
[19/Feb/2010:13:12:49 -0500] - listener got signaled
[19/Feb/2010:13:12:49 -0500] - get_ldapmessage_controls failed: 12 (Unavailable critical extension) (op=Abandon)
[19/Feb/2010:13:13:18 -0500] - activity on 68r
[19/Feb/2010:13:13:18 -0500] - read activity on 68
[19/Feb/2010:13:13:18 -0500] - conn 98 activity level = 3
[19/Feb/2010:13:13:18 -0500] - listener got signaled
[19/Feb/2010:13:13:18 -0500] Persistent Search - conn=98 op=2 The operation has been abandoned
[19/Feb/2010:13:13:18 -0500] - conn 98 turbo rank = 3 out of 5 conns
[19/Feb/2010:13:13:18 -0500] Persistent Search - conn=98 op=2 Releasing the connection and operation
[19/Feb/2010:13:13:18 -0500] - sasl_io_cleanup for connection 0
[19/Feb/2010:13:13:18 -0500] - sasl_pop_IO_layer: no SASL IO layer
[19/Feb/2010:13:13:18 -0500] - listener got signaled

###### NOT WORKING LOG ######
[19/Feb/2010:13:14:33 -0500] - activity on 68r
[19/Feb/2010:13:14:33 -0500] - read activity on 68
[19/Feb/2010:13:14:33 -0500] - conn 99 activity level = 2
[19/Feb/2010:13:14:33 -0500] - conn 99 turbo rank = 3 out of 5 conns
[19/Feb/2010:13:14:34 -0500] - get_ldapmessage_controls failed: 12 (Unavailable critical extension) (op=Abandon)
[19/Feb/2010:13:14:35 -0500] - listener got signaled
[19/Feb/2010:13:14:39 -0500] - activity on 68r
[19/Feb/2010:13:14:39 -0500] - read activity on 68
[19/Feb/2010:13:14:39 -0500] - conn 99 activity level = 2
[19/Feb/2010:13:14:39 -0500] - conn 99 turbo rank = 3 out of 5 conns
[19/Feb/2010:13:14:39 -0500] Persistent Search - conn=99 op=2 The operation has been abandoned
[19/Feb/2010:13:14:39 -0500] Persistent Search - conn=99 op=2 Releasing the connection and operation
[19/Feb/2010:13:14:39 -0500] - listener got signaled



On Thu, Feb 18, 2010 at 8:14 PM, Rich Megginson <rmeggins@redhat.com> wrote:
Chun Tat David Chu wrote:
> Hi All,
>
> I am running 389 DS version 1.2.5, build number 2010.012.2034 on RHEL 5.2.
>
> I have a problem that slapd didn't close a connection and eventually
> get into a CLOSE_WAIT state after my JAVA application exit.
>
> The scenario only happen when my application registers a
> NamingListener via the JAVA JNDI (Java Naming Directory Interface).  I
> believe the NamingListener is equivalent to the Persistent Search.
> This problem doesn't exist if I don't use the JNDI NamingListener
> capability.
>
> From my understanding, I did everything correctly in my application.
> I create a context, add a listener, do some stuffs, remove the
> listener and then close the context.
>
> One thing I notice is that in the slapd's error log, I see the
> following...
> "-get_ldapmessage_controls failed: 12 (Unavailable critical extension)
> (op=Abandon)".
> This message prints out right after I remove the listener and before
> my application closes the context.
>
> The closest bug report I found is this and it said the problem has
> been resolved.
> https://bugzilla.redhat.com/show_bug.cgi?id=450575
>
> At this point, I'm clueless.  :-(
> Can someone help me or give me some recommendation that I could try?
It looks as though the app is sending an LDAP Abandon request with
controls that are marked as critical, and the server does not support
them.  This is standard LDAPv3 behavior.  I'm not sure why it is using
Abandon, perhaps to Abandon any outstanding search or other requests
that have not completed.  Some things to check:
* see if you can specify that no controls are to be sent with the
Abandon request
* if you wait for all of the operations and results to be read by your
app, JNDI might notice this and just do an Unbind instead of an Abandon.
>
> I will attach my JAVA JNDI replicator along with this e-mail.  You
> will need to modify 2-3 lines of code to get it running in your
> environment.  Search for "MODIFY ME" and that should be the lines that
> you need to modify.
>
> Thanks!
>
> David
>
>
> ------------------------------------------------------------------------
>
> --
> 389 users mailing list
> 389-users@lists.fedoraproject.org
> https://admin.fedoraproject.org/mailman/listinfo/389-users

--
389 users mailing list
389-users@lists.fedoraproject.org
https://admin.fedoraproject.org/mailman/listinfo/389-users