(WTP SAR could cause an orphan). A dump statement would answer this problem, of course.
Aarno
On 2.3.2005, at 14:36, Jonathan Houser wrote:
Aarno,
Definitely ! Not to mention that our smsc, based on Kannel leaks fds, too. I do want to know is this a
different thing.
Update on the CLOSE_WAITs with the older (November) code. After 16 hours there's 20 stuck CLOSE_WAITs. So there's still a (tiny) leak. I'm going to try the January code next as Alex requested on Mantis.
Jonathan, are you using SAR ? Have you user-agent header in this event ?
Do you mean WAP or SMS SAR? I'm sure we're using WAP SAR for some of the files these people download. And I know for certain that every MMS Notification is two parts. However we're not doing 400+ MMS messages in 45 minutes on this box, which is how many CLOSE_WAITs I had when I reverted yesterday.
Off to get the older Kannel code from January and trial that!
Jon
