Turn off IPV6_USE_MIN_MTU?

Mukund Sivaraman muks at isc.org
Wed Oct 12 11:39:27 UTC 2016


Shane asked me to reply to this email about the current state of
things. As of the current releases of BIND, the following occurs:

For TCP over IPv6, named calls:

  (a) setsockopt(fd, IPPROTO_IPV6, IPV6_USE_MIN_MTU, ...) to 1

  (b) setsockopt(fd, IPPROTO_TCP, TCP_MAXSEG, ...) to (1280 - sizeof(tcp_header) - sizeof(ipv6_header)).

(b) was added later because in the case of TCP over IPv6 on FreeBSD, TCP
segments were larger than 1280 and they were getting fragmented.

For UDP over IPv6, named sets IPV6_USE_MIN_MTU = 1 (i.e., cap at 1280)
only when its reply message + sizeof(udp_header) + sizeof(ipv6_header) +
sizeof(ipv4_header[encapsulated]) cannot be sent within a 1500 bytes
PDU. So it sends out UDP packets upto 1500 bytes without fragmenting
them.

		Mukund
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 801 bytes
Desc: not available
URL: <https://lists.isc.org/pipermail/bind-workers/attachments/20161012/74f6f710/attachment.bin>


More information about the bind-workers mailing list