We are seeing very similar issues with ospf through our brocade. It is kind of like every so often a route goes stale and the brocade ignores it. I don’t have to kill upstream, Just change something in the /29 OSPF path and everything wakes up.
Have a Brocade Core and 3- 1072 distribution routers connected to Brocade onto different network segments. One Dist router has no issue getting to all sections but another will not be able to ping one or a few /29 or /30 even though all devices have the same number of routes. Quick change of a ospf route or flushing the Brocade OSPF records on the Dist router port clears the issue. Steve Barnes Wireless Ops Manager NLBC.com -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Dennis Burgess via Mikrotik-users Sent: Saturday, January 13, 2018 9:49 AM To: [email protected]; Mikrotik Users <[email protected]> Subject: Re: [Mikrotik Users] CCR 1072 BGP trivia for your Friday Night Sounds like an internal routing issue on the 1032 to me. And/or something up with your firewall. I would have to take a look. Dennis Burgess www.linktechs.net – 314-735-0270 x103 – [email protected] -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of mike.lyon--- via Mikrotik-users Sent: Friday, January 12, 2018 7:18 PM To: [email protected] Cc: Mikrotik Users <[email protected]> Subject: Re: [Mikrotik Users] CCR 1072 BGP trivia for your Friday Night No, i am only advertising the two /22s to my upstreams. Verified that they are only recieving the two /22s. I should also mention that the CCR 1072 connects to a CCR 1036 and the various /30s and /29s hang off the 1036. There is a 10.x /30 between the two routers. The 1036 has a default route pointing to the 1072. The 1072 has two staric routes routing the two /22s to the 1036. It would appear the route is getting dropped at the 1072. Joe Bob Random /30 will report that he is unable to browse the internet at one moment, traceroute stops at the 1072. Drop one of the two BGP upstreams (doesnt matter which one) and Joe Bob Random /30 begins to work again. Thanks, Mike > On Jan 12, 2018, at 17:08, <[email protected]> > <[email protected]> wrote: > > Are you actually advertising /29's and /30's to your providers? > > > Also, what exactly do you mean by dropping /30's? Like the packets are being > dropped to those ip's, or the /30's are being discarded, the routes are > disappearing? > > -----Original Message----- > From: [email protected] > [mailto:[email protected]] On Behalf Of mike.lyon--- via > Mikrotik-users > Sent: Friday, January 12, 2018 5:54 PM > To: Mikrotik Users <[email protected]> > Subject: [Mikrotik Users] CCR 1072 BGP trivia for your Friday Night > > Howdy! > > I have an interesting issue that i can’t seem to figure out... > > Quick overview: > > -1x CCR 1072 running 6.40.1 > -2 upstream providers, taking full tables from both to the 1072. > -2x public /22s behind the 1072 and am advertising both /22s to my upstreams. > -The /22s are chopped up into various /29s and /30s. > > Out of the blue, the 1072 will randomly drop various /30s (havent seen any > dropped /29s) out of those 2 /22s that im advertising upstream. > > If i drop one of either of my upstreams , the problem goes away and > everything is swell. > > Any ideas or thoughts on what may be causing this? > > Thank You, > Mike > > _______________________________________________ > Mikrotik-users mailing list > [email protected] > http://lists.wispa.org/mailman/listinfo/mikrotik-users > > _______________________________________________ Mikrotik-users mailing list [email protected] http://lists.wispa.org/mailman/listinfo/mikrotik-users _______________________________________________ Mikrotik-users mailing list [email protected] http://lists.wispa.org/mailman/listinfo/mikrotik-users _______________________________________________ Mikrotik-users mailing list [email protected] http://lists.wispa.org/mailman/listinfo/mikrotik-users
