[quagga-dev 4197] Re: possible ospfd bug: in state ExStart with routers that are neither DR nor BDR

Paul Jakma paul at clubi.ie
Fri Jun 30 02:12:45 BST 2006


On Thu, 29 Jun 2006, Andrew J. Schorr wrote:

> And when the T3 comes back up:
>
> 2006/06/29 13:39:30 OSPF: AdjChg: Nbr 192.168.21.30 on eth0.169:192.168.169.77: Down -> Init
> 2006/06/29 13:39:30 OSPF: AdjChg: Nbr 192.168.21.35 on eth0.169:192.168.169.77: Down -> Init
> 2006/06/29 13:39:30 OSPF: AdjChg: Nbr 192.168.21.30 on eth0.169:192.168.169.77: Init -> ExStart

Ok, how did /that/ happen. Init->ExStart (can only be 
twoway_received) /before/ any election? And once in ExStart there's 
no escape, which is strange because:

> 2006/06/29 13:39:30 OSPF: DR-Election[1st]: Backup 192.168.169.78
> 2006/06/29 13:39:30 OSPF: DR-Election[1st]: DR     192.168.169.79

Should have raised AdjOK? which should have set 30 back to 2-Way - 
but didn't.

> 2006/06/29 13:39:30 OSPF: AdjChg: Nbr 192.168.21.35 on eth0.169:192.168.169.77: Init -> ExStart

And 35 does the same. Odd.

And I've just read your followup email.. :)

> So it seems as if the two routers move to state ExStart prior to 
> the DR election (i.e. ti35 still thinks it's the DR).  Then the new 
> election occurs, and ti35 adjusts the state to 2-Way, but ti77 
> never makes the adjustment.  Is it proper that ti35 & ti77 move to 
> ExStart prior to the DR election?

No.

Nor is it proper that an election leaves them in ExStart when we 
don't consider them to be either DR/BDR (for Broadcast). AdjOK? 
should have been raised against every neighbour, and should have 
adjusted them back to 2-Way.

regards,
-- 
Paul Jakma	paul at clubi.ie	paul at jakma.org	Key ID: 64A2FF6A
Fortune:
"Irrationality is the square root of all evil"
  		-- Douglas Hofstadter



More information about the Quagga-dev mailing list