* UDP MSS [not found] <AANLkTi=qOQ0Rmyp78NEE9qSGOTFvgTDFSbDEDCtzoQ04@mail.gmail.com> @ 2010-12-14 3:47 ` ratheesh k 2010-12-14 4:00 ` Jan Engelhardt 1 sibling, 0 replies; 10+ messages in thread From: ratheesh k @ 2010-12-14 3:47 UTC (permalink / raw) To: Netfilter mailing list In udp , there is no mss discovey.So , if we set DF bit in IP packet and suppose an intermediate router does have disabled icmp-reply (thru iptables ) and have less MTU than the sender. Then , router will drop the packet ,How client know to fragment the udp packet ? -Ratheesh ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: UDP MSS [not found] <AANLkTi=qOQ0Rmyp78NEE9qSGOTFvgTDFSbDEDCtzoQ04@mail.gmail.com> 2010-12-14 3:47 ` UDP MSS ratheesh k @ 2010-12-14 4:00 ` Jan Engelhardt 2010-12-14 6:01 ` ratheesh k 2010-12-14 10:00 ` ratheesh k 1 sibling, 2 replies; 10+ messages in thread From: Jan Engelhardt @ 2010-12-14 4:00 UTC (permalink / raw) To: ratheesh k; +Cc: Netfilter mailing list On Tuesday 2010-12-14 04:46, ratheesh k wrote: >In udp , there is no mss discovey.So , if we set DF bit in IP packet and >suppose an intermediate router does have disabled icmp-reply (thru iptables >) and have less MTU than the sender. > >Then , router will drop the packet ,How client know to fragment the udp >packet ? It won't, that's why ICMP should not be blocked. ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: UDP MSS 2010-12-14 4:00 ` Jan Engelhardt @ 2010-12-14 6:01 ` ratheesh k 2010-12-14 10:00 ` ratheesh k 1 sibling, 0 replies; 10+ messages in thread From: ratheesh k @ 2010-12-14 6:01 UTC (permalink / raw) To: Jan Engelhardt; +Cc: Netfilter mailing list On Tue, Dec 14, 2010 at 9:30 AM, Jan Engelhardt <jengelh@medozas.de> wrote: > On Tuesday 2010-12-14 04:46, ratheesh k wrote: > >>In udp , there is no mss discovey.So , if we set DF bit in IP packet and >>suppose an intermediate router does have disabled icmp-reply (thru iptables >>) and have less MTU than the sender. >> >>Then , router will drop the packet ,How client know to fragment the udp >>packet ? > > It won't, that's why ICMP should not be blocked. > What will be the default MSS .How is it determined ? -ratheesh ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: UDP MSS 2010-12-14 4:00 ` Jan Engelhardt 2010-12-14 6:01 ` ratheesh k @ 2010-12-14 10:00 ` ratheesh k 2010-12-14 12:21 ` John Haxby 1 sibling, 1 reply; 10+ messages in thread From: ratheesh k @ 2010-12-14 10:00 UTC (permalink / raw) To: Jan Engelhardt; +Cc: Netfilter mailing list How udp determines maximum sustainable MSS ? On Tue, Dec 14, 2010 at 9:30 AM, Jan Engelhardt <jengelh@medozas.de> wrote: > On Tuesday 2010-12-14 04:46, ratheesh k wrote: > >>In udp , there is no mss discovey.So , if we set DF bit in IP packet and >>suppose an intermediate router does have disabled icmp-reply (thru iptables >>) and have less MTU than the sender. >> >>Then , router will drop the packet ,How client know to fragment the udp >>packet ? > > It won't, that's why ICMP should not be blocked. > ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: UDP MSS 2010-12-14 10:00 ` ratheesh k @ 2010-12-14 12:21 ` John Haxby 2010-12-14 13:40 ` ratheesh k 0 siblings, 1 reply; 10+ messages in thread From: John Haxby @ 2010-12-14 12:21 UTC (permalink / raw) To: ratheesh k; +Cc: Jan Engelhardt, Netfilter mailing list On 14/12/10 10:00, ratheesh k wrote: > How udp determines maximum sustainable MSS ? Unless I'm much mistaken, the concept doesn't apply. UDP doesn't have an MSS and "maximum sustainable MSS" would imply a connection that isn't there are well. jch ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: UDP MSS 2010-12-14 12:21 ` John Haxby @ 2010-12-14 13:40 ` ratheesh k 2010-12-14 13:45 ` Jan Engelhardt 0 siblings, 1 reply; 10+ messages in thread From: ratheesh k @ 2010-12-14 13:40 UTC (permalink / raw) To: John Haxby; +Cc: Jan Engelhardt, Netfilter mailing list On Tue, Dec 14, 2010 at 5:51 PM, John Haxby <john.haxby@oracle.com> wrote: > On 14/12/10 10:00, ratheesh k wrote: >> >> How udp determines maximum sustainable MSS ? > > Unless I'm much mistaken, the concept doesn't apply. UDP doesn't have an > MSS and "maximum sustainable MSS" would imply a connection that isn't there > are well. > > jch > > Coool .So udp fragmentation is done by ip layer.no mss. -Ratheesh ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: UDP MSS 2010-12-14 13:40 ` ratheesh k @ 2010-12-14 13:45 ` Jan Engelhardt 2010-12-14 14:18 ` ratheesh k 0 siblings, 1 reply; 10+ messages in thread From: Jan Engelhardt @ 2010-12-14 13:45 UTC (permalink / raw) To: ratheesh k; +Cc: John Haxby, Netfilter mailing list On Tuesday 2010-12-14 14:40, ratheesh k wrote: >On Tue, Dec 14, 2010 at 5:51 PM, John Haxby <john.haxby@oracle.com> wrote: >> On 14/12/10 10:00, ratheesh k wrote: >>> >>> How udp determines maximum sustainable MSS ? >> >> Unless I'm much mistaken, the concept doesn't apply. UDP doesn't have an >> MSS and "maximum sustainable MSS" would imply a connection that isn't there >> are well. >> >> jch >> >> > >Coool .So udp fragmentation is done by ip layer.no mss. Fragmentation is always done by the IP layer, even with protocols like TCP. You must mix up fragmentation and segmentation. ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: UDP MSS 2010-12-14 13:45 ` Jan Engelhardt @ 2010-12-14 14:18 ` ratheesh k 2010-12-14 14:35 ` Jan Engelhardt 0 siblings, 1 reply; 10+ messages in thread From: ratheesh k @ 2010-12-14 14:18 UTC (permalink / raw) To: Jan Engelhardt; +Cc: John Haxby, Netfilter mailing list On Tue, Dec 14, 2010 at 7:15 PM, Jan Engelhardt <jengelh@medozas.de> wrote: > Fragmentation is always done by the IP layer, even with > protocols like TCP. > You must mix up fragmentation and segmentation. Jan, what will be normal size of udp segments ? -Ratheesh ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: UDP MSS 2010-12-14 14:18 ` ratheesh k @ 2010-12-14 14:35 ` Jan Engelhardt 2010-12-14 14:57 ` ratheesh k 0 siblings, 1 reply; 10+ messages in thread From: Jan Engelhardt @ 2010-12-14 14:35 UTC (permalink / raw) To: ratheesh k; +Cc: John Haxby, Netfilter mailing list On Tuesday 2010-12-14 15:18, ratheesh k wrote: >On Tue, Dec 14, 2010 at 7:15 PM, Jan Engelhardt <jengelh@medozas.de> wrote: > >> Fragmentation is always done by the IP layer, even with >> protocols like TCP. >> You must mix up fragmentation and segmentation. > >Jan, > > what will be normal size of udp segments ? Counterquestion because yours does not seem to make sense: what is a normal size for TCP? ^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: UDP MSS 2010-12-14 14:35 ` Jan Engelhardt @ 2010-12-14 14:57 ` ratheesh k 0 siblings, 0 replies; 10+ messages in thread From: ratheesh k @ 2010-12-14 14:57 UTC (permalink / raw) To: Jan Engelhardt; +Cc: John Haxby, Netfilter mailing list On Tue, Dec 14, 2010 at 8:05 PM, Jan Engelhardt <jengelh@medozas.de> wrote: > Counterquestion because yours does not seem to make sense: what is a > normal size for TCP? > Jan, I got it. Thanks a ton. -Ratheesh ^ permalink raw reply [flat|nested] 10+ messages in thread
end of thread, other threads:[~2010-12-14 14:57 UTC | newest]
Thread overview: 10+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
[not found] <AANLkTi=qOQ0Rmyp78NEE9qSGOTFvgTDFSbDEDCtzoQ04@mail.gmail.com>
2010-12-14 3:47 ` UDP MSS ratheesh k
2010-12-14 4:00 ` Jan Engelhardt
2010-12-14 6:01 ` ratheesh k
2010-12-14 10:00 ` ratheesh k
2010-12-14 12:21 ` John Haxby
2010-12-14 13:40 ` ratheesh k
2010-12-14 13:45 ` Jan Engelhardt
2010-12-14 14:18 ` ratheesh k
2010-12-14 14:35 ` Jan Engelhardt
2010-12-14 14:57 ` ratheesh k
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.