On 10/04/06, Claudio Jeker <[EMAIL PROTECTED]> wrote:
>
> On Mon, Apr 10, 2006 at 06:56:33PM +0100, tony sarendal wrote:
> > On 10/04/06, Henning Brauer <[EMAIL PROTECTED]> wrote:
> > >
> > > * tony sarendal <[EMAIL PROTECTED]> [2006-04-10 19:04]:
> > > > On 10/04/06, Henning Brauer <[EMAIL PROTECTED]> wrote:
> > > > >
> > > > > * tony sarendal <[EMAIL PROTECTED]> [2006-04-08 00:51]:
> > > > > > It looks like bgpd has a problem with validating nexthop on new
> > > > > interfaces
> > > > > > when they are created.
> > > > > > A flap of the interface or restarting bgpd makes nexthop
> validate.
> > > > > > I have only tested with vlan interfaces.
> > > > >
> > > > > bizarre. I was able to see - let's call it "something odd" when I
> > > tried
> > > > > to reproduce that from home earlier this morning. Now I am
> completly
> > > > > unable to reproduce - it works just like it should. the
> RTM_IFANNOUNCE
> > > > > message that tells us about the new interface is directly followed
> by
> > > a
> > > > > RTM_IFINFO one giving us linkstate and the like and thus
> everything
> > > > > gets set allright.
> > > > > please:
> > > > > -run bgpd -d while doing this, and show output
> > > > > -run 'route monitor' as well
> > >
> > > > I run "ifconfig vlan26 create" and route monitor outputs:
> > >
> > > creating it manually is not needed.
> > >
> > > > got message of size 24 on Mon Apr 10 17:03:36 2006
> > > > RTM_IFANNOUNCE: iface arrival/departure: len 24, if# 13, name
> vlan26,
> > > what:
> > > > arrival
> > > >
> > > >
> > > >
> > > > I run " ifconfig vlan26 vlan 26 vlandev pcn1" and route monitor
> outputs:
> > > >
> > > > got message of size 96 on Mon Apr 10 17:04:02 2006
> > > > RTM_NEWADDR: address being added to iface: len 96, metric 0, flags:
> > > > sockaddrs: <NETMASK,IFP,IFA>
> > > > ffff:ffff:ffff:ffff:: 00:0c:29:25:74:9f
> fe80::20c:29ff:fe25:749f%vlan26
> > > > got message of size 124 on Mon Apr 10 17:04:02 2006
> > > > RTM_ADD: Add Route: len 124, pid: 0, seq 0, errno 0,
> > > flags:<UP,HOST,LLINFO>
> > > > locks: inits:
> > > > sockaddrs: <DST,GATEWAY>
> > > > fe80::20c:29ff:fe25:749f%vlan26 00:0c:29:25:74:9f
> > > > got message of size 188 on Mon Apr 10 17:04:02 2006
> > > > RTM_ADD: Add Route: len 188, pid: 0, seq 0, errno 0,
> > > flags:<UP,DONE,CLONING>
> > > > locks: inits:
> > > > sockaddrs: <DST,GATEWAY,NETMASK,IFP,IFA>
> > > > fe80::%vlan26 link#13 (255)
> > > > Qff.ff.ff.ff.ff.ff.ff.ff.ff.ff.ff.ff.ff.ff00:0c:29:25:
> > > > 74:9f fe80::20c:29ff:fe25:749f%vlan26
> > >
> > > wow. if you use the create, assign vlan/vlandev sequence ther eis no
> > > RTM_IFINFO! that's why we don't have no flags and stuff.
> > >
> > > > I now flap the interface:
> > > >
> > > > cr203-STO# ifconfig vlan26 down; ifconfig vlan26 up
> > > >
> > > > bgpd -d reports:
> > > >
> > > > nexthop 10.1.1.38 now valid: directly connected
> > >
> > > yep, bacuse that causes an RTM_IFINFO and thuwe we learn link state
> and
> > > flags and stuff
> > >
> > > bizarre. I'll try to look closer when I find some time
> > > (I do have a small diff that solves it, fetching teh info from sysctl,
> > > but I really want to find out what goes wrong here. fetching via
> sysctl
> > > should NOT be needed.)
> >
> >
> >
> > Cool. No need to rush it for me, I am in a testing environment and have
> time
> > to wait for a fix, and solid fixes are nice.
> >
>
> Could you try a different interface (something else than pcn(4)). Could be
> a driver specific issue (e.g. em(4) creates the RTM_IFINFO)
>
> --
> :wq Claudio
>
>
I'll see if I can test it on boxes with dc(4) instead, although I belive you
are correct.
It works just as it should with gif(4) interfaces when adding/removing them.
/Tony
--
Tony Sarendal - [EMAIL PROTECTED]
IP/Unix
-= The scorpion replied,
"I couldn't help it, it's my nature" =-
