I see memory increased just right after several minutes and then it is stable so there can be leaks.
I'm using windows.
Hello nick,
I tried your page last night. I had to move the content
to a facelet (I kept getting a "redirect error" when I tried
the jsp ). Otherwise, the test was unchanged.
I did not see the node count leak increase in this case. I will try
a longer test today to be sure. In my application I get a very reliabe one node/poll
increment.
In my case, obviously, there is a bit more data and interaction with the
backing bean (in particular, the bean provides a value for renderList).
I will try to attach a few graphs to the jira showing what I am seeing. The slope
of the "leak" case is very obvious, compared with the run where onsubmit was removed.
In my case, the poller is rerendering a rich:datatable and a few smaller panels.
I have found that changing the the number of rows in the table will change the slope of
the plot. I.e., if you increase the row count, you get a sharper slope. This hints, of course, that the objects that comprise the rows are not being cleaned up. Exactly why this would be exposed by the use of onsubmit on the poll is a mystery to me....
PdP
PdP,
So we need the full page code to reproduce and investigate the problem at our side. Can you please prepare little sample? You can attach it to JIRA issue or send it to me to nbelaevski at exadel dot com. Thank you in advance!
Hi,
I suggest that we move to this topic: http://www.jboss.com/index.html?module=bb&op=viewtopic&t=127124&postdays=0&postorder=asc&start=10
kshenes,
Can you post full page code? I need to know what is "statsGrid,status,stausIcon,master,messages". And again, let's clarify what browsers are affected by this problem? One more question to investigate is whether it is a particular poll issue or the issue of another component multiplied by multiple page updates made by poll?