Date: Thu, 14 Feb 2013 12:29:29 +0100 (CET) From: Wojciech Puchar <wojtek@wojtek.tensor.gdynia.pl> To: "Teske, Devin" <Devin.Teske@fisglobal.com> Cc: "freebsd-hackers@freebsd.org" <freebsd-hackers@freebsd.org> Subject: RE: rsh/rlogin strange behavior Message-ID: <alpine.BSF.2.00.1302141228340.2202@wojtek.tensor.gdynia.pl> In-Reply-To: <13CA24D6AB415D428143D44749F57D7201EA93B6@ltcfiswmsgmb21> References: <alpine.BSF.2.00.1302131837470.1552@wojtek.tensor.gdynia.pl>, <alpine.BSF.2.00.1302141135570.1901@wojtek.tensor.gdynia.pl> <13CA24D6AB415D428143D44749F57D7201EA93B6@ltcfiswmsgmb21>
next in thread | previous in thread | raw e-mail | index | archive | help
> > repeat 100 rsh <host> date > > HINT: Set yourself up in /etc/hosts.equiv on <host> for password-less entry > > Repeat about 5 or 6 times and then eventually the connection will hang and you won't be able to make more connections for some time. > > Next step? Execute "netstat -an | less" and look for oddities (like a mass pile of FIN_WAIT_2 connections). > > In our case (ymmv) the final ACK was not being sent leaving the client side stacking up a bunch of connections that take msl.timeout time to expire (iirc). If I do remember correctly the problem happened when the server was using an em(4) driver. > > Our ultimate solution was to either switch critical servers to fxp(4) based hardware or roll entire sites over to using key-based SSH (which may work for you -- have you thought about giving ssh-keygen a try? that is, if you're using rsh for the convenience of password-less entry via hosts.equiv for example). > -- it is FreeBSD 9, em or re or bge hardware but rlogin goes over tun(4) interface. in the same time rcp works fine even for gigabyte file. any more ideas?
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?alpine.BSF.2.00.1302141228340.2202>