From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Jim Fleming" Subject: Re: disturbing MTU experiment Date: Tue, 8 Apr 2003 13:46:11 -0500 Sender: netfilter-devel-admin@lists.netfilter.org Message-ID: <007601c2fdff$20a81790$8500a8c0@repligate> References: <20030408154101.13919.89741.Mailman@kashyyyk> <16018.64600.149601.372694@isis.cs3-inc.com> Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Return-path: To: "Don Cohen" , Errors-To: netfilter-devel-admin@lists.netfilter.org List-Help: List-Post: List-Subscribe: , List-Unsubscribe: , List-Archive: List-Id: netfilter-devel.vger.kernel.org Are you on the AM Internet or the FM InterNAT ? 128-bit DNS AAAA Record Flag Day Formats 2003:[IPv4]:[SDLL.OFFF.FFFF.TTTT]:[64-bit IPv8 or IPv16 Persistent Address] [YMDD]:[IPv4]:[SDLL.OFFF.FFFF.TTTT]:[64-bit IPv8 or IPv16 Persistent Address] 1-bit to set the Reserved/Spare ("AM/FM") bit in Fragment Offset [S] 1-bit to set the Don't Fragment (DF) bit [D] 2-bits to select 1 of 4 common TTL values (255, 128, 32, 8) [LL] 1-bit for Options Control [O] 7-bits to set the Identification Field(dst) [FFFFFFF] 4-bits to set the TOS(dst) Field [TTTT] Default SDLL.OFFF.FFFF.TTTT = 0000.0000.0000.0000 FFF.FFFF.TTTT = GGG.SSSS.SSSS http://www.ntia.doc.gov/ntiahome/domainname/130dftmail/unir.txt IPv8 0QQQQGGGSSSSSSSS[32-bits][Port] IPv16 0QQQQGGGSSSSSSSS[32-bits][Port] 1AAAAAAAAAAAAAAA[32-bits][Port] A...A=ASN=32769...65535 Jim Fleming http://www.IPv8.info ----- Original Message ----- From: "Don Cohen" To: Sent: Tuesday, April 08, 2003 11:44 AM Subject: disturbing MTU experiment > > (Let me know if there's a more appropriate place to send this.) > > My understanding of IPv4 is that > - the forwarding path between two machines might have links of > different MTUs, > - when an IP datagram marked DF is too big for one of those links, the > machine that would otherwise forward it is supposed to return an > ICMP unreachable reply > - the sending machine is supposed to react to that ICMP reply, either > by sending a smaller datagram or by allowing fragmentation or of > course, giving up. > > The following experiment ought to exhibit this behavior: > > internet --- eth0:linux_firewall:eth1 --- client > > - on linux firewall do > ifconfig eth1 mtu 1400 > - on client machine try to use the internet, e.g., run a web browser > > When I do this I find, to my dismay, that MANY sites don't work! > A tcpdump on the firewall shows the web server in the internet sending > a DF packet of size 1500, the firewall sending the ICMP reply, and the > server ignoring it, i.e., resending the large packet over and over. > > 1. Is there something wrong with my experiment or is a large part of > the internet really broken in this way? Do others out there see the > same thing? > > If I am correct, then > > 2. What's the cause of this breakage? Are servers filtering ICMP due > to attacks? Are they behind firewalls that don't know how to forward > the ICMP packets? Are ISPs filtering the ICMP packets? Are the > servers themselves running broken network code? If so, which are the > OS's with this defect? > > 3. Are links with MTU < 1500 are extremely rare in the internet? > The fact that the internet mostly seems to work would suggest that. > Surely this has not always been the case. When did it happen? > Wouldn't this break various tunneling protocols? At least those that > try to encapsulate one packet in one IP packet. Maybe those are also > not much used? > > 4. What can be done about it? For instance, what would happen if > the machines that send ICMP replies were to go ahead and fragment > the DF packets (perhaps instead of sending out a second ICMP to the > same offender)? >