TIME_WAIT does not exclusively happen on the server. Stephen Johnston
On Wed, Apr 7, 2010 at 5:44 PM, Jay Paroline <[email protected]> wrote: > This may be due to a basic misunderstanding on my part, but in that > case, if I see lots of entries like this from netstat -tnap running on > the client side, what does it mean? > > tcp 0 0 172.16.0.15:35672 > 172.16.0.64:11211 TIME_WAIT - > > 172.16.0.15 is the server I ran netstat -tnap on (the client in this > case) and 172.16.0.64:11211 is the memcache server. > > Jay > > On Apr 7, 4:19 pm, Brian Moon <[email protected]> wrote: > > > @Tyn, I'll look into sharing a class instance, didn't know that was > > > possible in APC. > > > > It is not. Well, you can recreate a class, but the connection will not > > be the same. It will have to recreate that resource. > > > > > @Brian Moon, I know the memcache servers can handle having all those > > > incoming connections, I'm just trying to reduce the number of > > > ephemeral ports we need on each web server to handle all the outgoing > > > connections. My understanding is that TIME_WAIT uses up an ephemeral > > > port the same as ESTABLISHED. > > > > TIME_WAIT is a host side state, not client side. > > > > "When you close a socket, the server goes into a TIME_WAIT state, just > > to be really really sure that all the data has gone through." > > > > http://www.unixguide.net/network/socketfaq/2.7.shtml > > > > Brian. > -- To unsubscribe, reply using "remove me" as the subject.
