Ondrej - I agree with your assessment (A) is the way to go. Now we need to determine if Quagga has fixed this issue between .99.22 and .99.24
thanks! donald On Thu, Jun 18, 2015 at 9:20 AM, Ondrej Zajicek <[email protected]> wrote: > On Thu, Jun 18, 2015 at 08:39:28AM -0400, Donald Sharp wrote: > > Ondrej - > > > > Reading 10.9 of RFC2328 leads me to two interpretations of it: > > > > (A) Your interpretation, as stated below. > > (B) When the originating router receives a LSUPD, it should truncate it's > > list and immediately send a new LSREQ packet for the remaining items. > > Unfortunately there is no sequence number in LSREQ / LSUPD, so it is not > possible to determine relation between LSREQs and LSUPDs. The approach > (B) is problematic, because it may lead to request storms: > > X sends LSREQ1 for 1-60 > Y receives LSREQ1, sends LSUPD1 for 1-20, LSUPD2 with 21-40, LSUPD3 with > 41-60 > X receives LSUPD1, responds with LSREQ2 for 21-80 > X receives LSUPD2, responds with LSREQ3 for 41-100 > X receives LSUPD3, responds with LSREQ4 for 61-120 > Y receives LSREQ2, sends LSUPD4 for 21-40, LSUPD5 for 41-60, LSUPD6 for > 61-80 > Y receives LSREQ3, sends LSUPD7 for 41-60, LSUPD8 for 61-80, LSUPD9 for > 81-100 > Y receives LSREQ4, sends LSUPD10 for 61-80, LSUPD11 for 81-100, LSUPD12 > for 101-120 > ... > > In this example every LSA is retransmitted three times, but in the worst > case scenario > the incoming LSUPD would bring only one new LSA. > > The approach (B) could work properly but in that case the behavior of the > other side (reaction to LSREQ) also have to be modified - there are two > approachs: > > (C) When router receives a LSREQ, it immediately send multiple LSUPD > packets for all LSAs (as mentioned in my original post) > > (D) When router receives a LSREQ, it immediately send just one LSUPD > with just enough LSAs to fit to one LSUPD. > > Approach (A) would work with (C), approach (B) would work with (D), > but mixing (A)/(D) or (B)/(C) would lead > > > Having said that, 0.99.22 of Quagga is just over 2 years old now and > from a > > quick glance at the logs I see ~70 commits into OSPF alone in that time > > frame. Is there an easy way to upgrade to a newer version of Quagga? > > Unfortunately i just reposting the issue bringed on at BIRD mailing list. > I have logs from this case but i currently have no running Quagga / BIRD > lab to test this. > > -- > Elen sila lumenn' omentielvo > > Ondrej 'Santiago' Zajicek (email: [email protected]) > OpenPGP encrypted e-mails preferred (KeyID 0x11DEADC3, wwwkeys.pgp.net) > "To err is human -- to blame it on a computer is even more so." >
_______________________________________________ Quagga-dev mailing list [email protected] https://lists.quagga.net/mailman/listinfo/quagga-dev
