The nav refresh was indeed the culprit of the recent memory leak (thanks Scott for helping track it down). The way I was calling the refresh resulted in a closure that did not quite behave the way I thought it would. I wasn't able to get it to quite do what I wanted tonight, so this patch simplifies the way we are doing it for now. It should behave the same as before, minus the memory leak, but in the near future I will try and get it to work the way I originally wanted. -j -------------- next part -------------- A non-text attachment was scrubbed... Name: s3-memory-leak.patch Type: text/x-patch Size: 4574 bytes Desc: not available URL: <http://listman.redhat.com/archives/ovirt-devel/attachments/20080610/9096af31/attachment.bin>
On Tue, Jun 10, 2008 at 01:15:39AM -0400, Jason Guiditta wrote:> The nav refresh was indeed the culprit of the recent memory leak (thanks > Scott for helping track it down). The way I was calling the refresh > resulted in a closure that did not quite behave the way I thought it > would. I wasn't able to get it to quite do what I wanted tonight, so > this patch simplifies the way we are doing it for now. It should behave > the same as before, minus the memory leak, but in the near future I will > try and get it to work the way I originally wanted. > > -jACK. I have committed this, seems to work fine in test (although I would have to watch longer to see for sure that the leak is gone). Take care, --Hugh
Jason Guiditta wrote:> The nav refresh was indeed the culprit of the recent memory leak (thanks > Scott for helping track it down). The way I was calling the refresh > resulted in a closure that did not quite behave the way I thought it > would. I wasn't able to get it to quite do what I wanted tonight, so > this patch simplifies the way we are doing it for now. It should behave > the same as before, minus the memory leak, but in the near future I will > try and get it to work the way I originally wanted. > > -j >Definitely an improvement. I'm still seeing one extra request per refresh period. In addition to the tree/fetch_json call, the top-level URL is being requested again. So, if you start at the dashboard, whatever pool you're looking at the full html response for the dashboard page is being requested. So we still have a leak, but it's much slower than before. Scott