Re: Fragment question again ..

Hi Gaurav

No, I think you are not correct, the router do not wait for the MF = 0 to
build the packet.

In an ideal network or LAN there is no reorder and in that ideal world
waiting for MF = 0 would indicate reassemble now

But we are not in an ideal world and if you have an easy solution to solve a
lot of packet drop why not do it, by implementing a reassemble timeout.
there is also an ICMP type 11 code 1 for fragment exceeded timeout on
reassembly.

*RFC 792:*
*
*

*If a host reassembling a fragmented datagram cannot complete the
      reassembly due to missing fragments **within** its time limit it
      discards the datagram, and it may send a time exceeded message.*

within == start || middle || end

You can also look into some code how fragment reassemble work:

http://www.linuxhq.com/kernel/v2.6/24/net/ipv4/ip_fragment.c

On Mon, Sep 27, 2010 at 7:45 AM, GAURAV MADAN <gauravmadan1177_at_gmail.com>wrote:

> HI Shiran
>
> This is not always correct ..
>
> Lets assume that fragment ( last fragment ) with MF bit set to 0
> reaches the final (dest) router before the 3.3 fragment in your case .
> In such a case ; router has already received a fragment with MF = 0
> and hence will not consider any other fragments coming with same
> Identifier bytes .
>
> I hope you are getting my ques .
>
> Thnx
> Happy !
>
> On Sun, Sep 26, 2010 at 11:00 PM, shiran guez <shiranp3_at_gmail.com> wrote:
> > The router that will frag the fragment will take the 1500 byte and will
> set
> > the offset for the new frag
> >
> >
> > so you will have
> >
> > 1 pkt 1500 with offset 0 from 5000
> > 2 pkt 1500 with offset 1501 from 5000
> > 3 pkt 1500 with offset 3001 from 5000--> now lets say that packet 3 is
> going
> > through another frag off 500 byte
> > 3.1 pkt 500 with offset 0 from 1500
> > 3.2 pkt 500 with offset 501 from 1500
> > 3.3 pkt 500 with offset 1001 from 1500
> > 4 pkt 500 with offset 4501 from 5000
> >
> > so the router will receive fragment 1 2 and now 3 he will see another
> > fragment and only after 3 is assemble he will be able to assemble 4
> >
> > On Sun, Sep 26, 2010 at 11:29 AM, Happy Singh <
> > happynetworkingsingh_at_gmail.com> wrote:
> >
> >> R1 -------------- R2 -------------- R3
> >> | |
> >> | |
> >> | |
> >> R4---R5----R6------R8-----------
> >>
> >>
> >> Hi All
> >>
> >> Last week I discussed a few points abt fragmntation in IP networks .Thnx
> >> for
> >> all help . Here is a interesting scenario which i wanted to clearify
> >>
> >> Let 5000 byte pkt is rx on R1 . R1 makes 4 fragments : 1500 , 1500 ,
> 1500 ,
> >> 500
> >>
> >> Let is sends Frag 1 ( offset 0 ) , Frag 3 ( offset 3001 ) , fragment 4 (
> >> offset 4501 ) to link1
> >> Let it sends Frag 2 ( offset 1501) to link 2
> >>
> >> Last fragment on link 1 will have MF = 0 ; others will have MF = 1 ( MF
> =
> >> more fragments )
> >>
> >> Now the pkt that goes via R4--R5--R6--R8 is required to be further
> >> fragmented ( due to even smaller MTU )
> >> Let it is broken as frag 1 ( offset 1501) ; frag 2 ( offset 2001) ; frag
> 3
> >> (
> >> offset 2501 ) : 4 sub-fragments of 500 bytes each .
> >> Now these will have MF = 1 and fragment with offset 2501 will have MF =
> 0
> >>
> >>
> >> Question is : AT assembling router (R3) ; we are getting 2 fragments
> with
> >> MF = 0 . This will cause problem because if fragment 4 ( offset 4501 )
> with
> >> MF = 0 is recieved before frag # ( offset 2501 ) then it might assume
> that
> >> since MF = 0 ; I am not expecting any other fragments .
> >>
> >> Please tell me how is this taken care of in routers ?
> >> I hope I have made my question clear ,
> >>
> >> Thnx
> >> Happy !
> >>
> >>
> >> Blogs and organic groups at http://www.ccie.net
> >>
> >> _______________________________________________________________________
> >> Subscription information may be found at:
> >> http://www.groupstudy.com/list/CCIELab.html
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >
> >
> > --
> > Shiran Guez
> > MCSE CCNP NCE1 JNCIA-ER CCIE #20572
> > http://cciep3.blogspot.com
> > http://www.linkedin.com/in/cciep3
> > http://twitter.com/cciep3
> >
> >
> > Blogs and organic groups at http://www.ccie.net
> >
> > _______________________________________________________________________
> > Subscription information may be found at:
> > http://www.groupstudy.com/list/CCIELab.html
> >
> >
> >
> >
> >
> >
> >
> >
>

-- 
Shiran Guez
MCSE CCNP NCE1 JNCIA-ER CCIE #20572
http://cciep3.blogspot.com
http://www.linkedin.com/in/cciep3
http://twitter.com/cciep3
Blogs and organic groups at http://www.ccie.net