QE engineer Vivek verified both the scenarios you outlined, with this
build and could see sessions resumed correctly (could not reproduce your
issue).
Lets see what is different in your environment.
a. Are you running this test with the clusterjsp application? If not,
could you deploy this app and run your test and let us know the results
(perhaps with screenshots)
b. Could you look into the logs and see if there are any exception stack
traces in any of the instances including DAS. ( particularly anything
suggesting pipes being closed)
c. Were you running your earlier test on a single machine or on
different machines (one instance on each machine) - test we ran was on
single machine? (could you also check to see if replication is happening
at all between the remaining two instances after the first instance is
killed/stopped.
d. Use the following logger setting to get more log information on
session replication - under Logger settings in the GUI,
add a property: name="com.sun.enterprise.ee.web.sessmgmt" value="FINE"
and another property : name="com.sun.enterprise.ee.web.sessmgmt.pipe"
value="FINE"
e. Could you ensure that the clocks on all machines involved are sync'd
up and they are all on the same subnet (I think they should be as
scenario a would not have worked otherwise).
Thanks
Shreedhar
glassfish_at_javadesktop.org wrote:
>> Just to confirm, you are not running this test under any load and can be
>> reproduced with a few sessions, right?
>>
>
> Right, I'm not running this teste under any loadAre
> And I can reproduce the second scenario every time I try.
>
> If you need more info, please, let me know.
>
>
> Thank you.
> [Message sent by forum member 'galmeida' (galmeida)]
>
> http://forums.java.net/jive/thread.jspa?messageID=357456
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe_at_glassfish.dev.java.net
> For additional commands, e-mail: users-help_at_glassfish.dev.java.net
>
>