Description of problem: The -M option on IPv4 ping controls setting of the don't fragment bit. The option 'do' is described as: do (prohibit fragmentation, even local one) IPv6 doesn't provide network based fragmentation, so the -M pmtu setting should only apply locally, the man page for ping implies it will, but it does not. The ability to control fragmentation with ping is very useful for network debugging esp in an enviroment with tunnels as is common with IPv6. Version-Release number of selected component (if applicable): iputils-20020927-22 How reproducible: Completely Steps to Reproduce: 1. ping6 -s 8000 -M do dualstack_host 2. ping -s 8000 -M do dualstack_hist 3. Actual results: [root@hypatia ~]# ping6 -s 8000 -M do goethe.freenode.net PING goethe.freenode.net(sterling.freenode.net) 8000 data bytes 8008 bytes from sterling.freenode.net: icmp_seq=1 ttl=58 time=292 ms 8008 bytes from sterling.freenode.net: icmp_seq=2 ttl=58 time=302 ms 8008 bytes from sterling.freenode.net: icmp_seq=3 ttl=58 time=293 ms 8008 bytes from sterling.freenode.net: icmp_seq=4 ttl=58 time=290 ms --- goethe.freenode.net ping statistics --- 5 packets transmitted, 4 received, 20% packet loss, time 4006ms rtt min/avg/max/mdev = 290.902/294.720/302.116/4.370 ms, pipe 2 Expected results: Some sort of ICMP error about fragmentation needed. Additional info:
An aside, but wouldn't tracepath6 be a better tool for debugging things like tunnel sizes?
(In reply to comment #1) > An aside, but wouldn't tracepath6 be a better tool for debugging things like > tunnel sizes? Yep. Doesn't mean ping shouldn't work as expected. ;)
(In reply to comment #0) > Description of problem: > > IPv6 doesn't provide network based fragmentation, so the -M pmtu setting should > only apply locally, the man page for ping implies it will, but it does not. > > The ability to control fragmentation with ping is very useful for network > debugging esp in an enviroment with tunnels as is common with IPv6. > Does this mean that it should work or should not, cos IPv6 doesn't have a capability for it?
Of course IPv6 has a capability for fragmentation--just end-host based, not router based fragmentation. This bug still exists in iputils-20070202-5.fc8. Please update the version to "devel". Thanks.
Why is this bug still sitting at need-info? :( I don't see what more information is needed.
Gregory Maxwell, since you are the submitter of this bug, could you please change the Version to devel and the Status to ASSIGNED or similar? Only the submitter, assignee, and a few others can make such changes. I tried clicking "I am providing the requested information for this bug." but it doesn't work for me.
(In reply to comment #0) > Version-Release number of selected component (if applicable): > iputils-20020927-22 The iputils-20020927-22 is from FC4. There is a test result of release iputils-20071127 of FC9 below. I tested also older releases (FC8) and works also fine. [jskala@...]$ ping6 -c 4 -s 8000 -M do fec0::f101:215:c5ff:fef3:e54f PING fec0::f101:215:c5ff:fef3:e54f(fec0::f101:215:c5ff:fef3:e54f) 8000 data bytes From fec0::f101:21c:25ff:fe76:6547 icmp_seq=1 Packet too big: mtu=1500 From fec0::f101:21c:25ff:fe76:6547 icmp_seq=1 Packet too big: mtu=1500 From fec0::f101:21c:25ff:fe76:6547 icmp_seq=1 Packet too big: mtu=1500 From fec0::f101:21c:25ff:fe76:6547 icmp_seq=1 Packet too big: mtu=1500 --- fec0::f101:215:c5ff:fef3:e54f ping statistics --- 0 packets transmitted, 0 received, +4 errors