Date: Wed, 16 Dec 1998 09:57:49 +0100 From: Dirk-Willem van Gulik <dirk.vangulik@jrc.it> To: freebsd-hackers@FreeBSD.ORG Subject: Re: EADDRINUSE rather than EADDRNOTAVAIL Message-ID: <3677760D.B7B7A4FA@jrc.it>
next in thread | raw e-mail | index | archive | help
Larry Baird wrote: > In article <367664AE.D50A3A75@jrc.it> you wrote: > : I've got a small database transaction client/server pair. It is > : to accomplish serialisation of a db shared by a couple of web > : servers on a handfull of machines. The backend is DD; typical > : transaction speeds are in the 50/client/second; and 1500/server/second. > Look at http://www.ibrado.com/sock-faq/ for the answer to your question. Nice plug, for a very good FAQ. However the question I am seeking to get answered is is a bit more complex; or my brain is just not into gear to spot it. On a busy client; after opening and closing a couple of 100 tcp/ip sockets in the AF family to the server on another machine I am getting EADDRINUSE as it seems to run out of its local ports. 1. now why is this. Surely a couple of hundred open should not be an issue. 2. and why do I not get something lik EADDRNOTAVAIL ? I mean the kernel should make sure that the port is not in use when I ask for one. 3. and what can I tune to 'fix' this. Or would this be caused by me shooting myself in the foot by increasing the FD_SETSIZE (for more FDs) and BSIZE=256 to get past the 101-254 hole in delayed send. Thanks, Dw. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3677760D.B7B7A4FA>