From mboxrd@z Thu Jan 1 00:00:00 1970 MIME-Version: 1.0 In-Reply-To: <7b531b6b4731db24073f6404b374c8eb@brasstown.quanstro.net> References: <203975f49e91f93a5b4df848ead8e5c2@kw.quanstro.net> <7b531b6b4731db24073f6404b374c8eb@brasstown.quanstro.net> Date: Thu, 22 Jul 2010 10:55:48 -0700 Message-ID: From: Akshat Kumar To: Fans of the OS Plan 9 from Bell Labs <9fans@9fans.net> Content-Type: text/plain; charset=ISO-8859-1 Subject: Re: [9fans] auth server with two NICs Topicbox-Message-UUID: 42661a40-ead6-11e9-9d60-3106f5b1d025 On Thu, Jul 22, 2010 at 10:30 AM, erik quanstrom wrote: > a bad network? I thought so at first, but if instead of using separate /net and /net.alt mountpoints for the two networks, I simply, as I said before, bind -b '#l1' /net bind -b '#I1' /net and start auth service, etc., afterwards (so that they start only on the external interface), then from testing last night, there is no such lag in drawterm or dropped packets on ping. now, it could just be the timing of things (like, when I tested, etc.)... but if I drawterm to a remote network, it's *much* faster, and pinging around to other things doesn't cause dropped packets. So, I'm lead to believe that the problem is in a configuration of two network interfaces, where the default is the internal network, which is bound to /net, and the external network is bound to /net.alt is there some reason or way that incoming calls could be confused, or it might take time for the server to find the right place to reply from? I dunno... but the performance is killing me.