Ok, I have run some other tests with mina-core-2.0.2 and this connection pattern is still there... It surely has to deal with Windows since, on Linux and Mac OS X there's no sign of it.
I am still trying to capture traffic going through any of these connection but it can't see any of it. Frankly I'm a bit concerned as we had a system failure weeks ago where Windows reported it was out of nonpaged pool, and it stopped being able to create new connections. On average netstat shows around 250-300 established connections of which some 80 are these "unknown" ones. It also has 30-40 TIME_WAIT ones. Since the app we're working on basically is just a webapp using mina and a DB, and being able to pinpoint db connections easily, do you have some ideas about what could be causing this and if we should worry about it (right now we do, maybe it's a well known Windows/NIO behavior that I haven't been able to locate). Thanks again, regards Francesco On Tue, Feb 8, 2011 at 11:33 AM, Francesco Vivoli <[email protected]>wrote: > we have migrated to 2.0.2 in our dev environment, I still have to run tests > in pre-production though. I'll follow up once I have done so and checked > whether this behavior persists. > > regards > F > > > On Tue, Feb 8, 2011 at 10:29 AM, Emmanuel Lecharny <[email protected]>wrote: > >> On 2/8/11 10:22 AM, Francesco Vivoli wrote: >> >>> Hi there >>> >>> I'm still in the process of assessing an application which make a fairly >>> heavy use of mina. It is packaged as a webapp deployed under tomcat. >>> >>> Using netstat I have noticed a strange (to me) connection pattern like >>> the >>> one attached below. I understand that this is very application specific >>> and >>> that it's hard to draw conclusions just looking at this trace, but this >>> application itself doesn't open internal connections this way: it is >>> basically a bunch of acceptors and connectors, so nothing that would >>> explain >>> a netstat output like this. Moreover, I have used a sniffer to inspect >>> traffic but I haven't captured any of it. Strange. >>> >>> Does anybody have ever noticed something like this? The server running >>> this >>> application has shown to sometimes to run out of resources (not being >>> able from >>> the system nonpaged pool and thus not being able to create any more >>> connections) and I'm trying to get to the cause. I have been monitoring >>> the >>> trend of these connections and they seem to enlarge and shrink but >>> definitely they show a steady growth. Still, I don't know who's opening >>> them. I hope you guys can help:) >>> >>> Btw, production is still running 2.0.0-M6. >>> >> >> First thing I would do would be to switch to MINA 2.0.2. 2.0.0-M6 is just >> totally outdated. >> >> >> -- >> Regards, >> Cordialement, >> Emmanuel Lécharny >> www.iktek.com >> >> >
