We’re testing vyos 1.5 to be a BNG replacement for our cisco gear. When we’ve configured OSPF with the following
set protocols ospf area 1.0.0.0 network ‘1.2.3.4/28’<ip’s redacted>
set protocols ospf interface bond0 passive disable
set protocols ospf interface bond0 priority ‘0’
set protocols ospf log-adjacency-changes
set protocols ospf passive-interface ‘default’
set protocols ospf redistribute connected
the ospf adjacencies almost immediately enter the loading DR and BDR, however it sits there for around 11 minutes before entering full, and seeming to work properly after that. This happens every OSPF restart, or device reboot.
The juniper devices we’re connecting to are showing full within about 10seconds, however we’re not receiving any routes from the vyos box at that time, once vyos shows full, all things start to ping.
We’ve fully checked MTU, firewall rules(even removing and allowing everything), and have a ton of logs, we haven’t tried updating to q2 yet, as q2 has some configuration differences that I have to report separately, (set firewall zone subscriber interface ‘l2tp*’ doesn’t work as a command).
Timers from vyos: Timer intervals configured, Hello 10s, Dead 40s, Wait 40s, Retransmit 5
Timers from junos: Hello: 10, Dead: 40, ReXmit: 5, Not Stub
You say after restart of service and device reboot.
Is this time you mention the full time or time after VyOS itself finished loading?
Because there is this issue I reported about 2 years ago:
And since you mention you use this box for BNG I assume there can be more than “a handful of routes” (or firewall rules) to get loaded?
If it is this case one thing you could try out (to verify) would be to just remove all but lets say 1-2 routes/firewall rules, reboot and see if it still takes about 11 minute for the box to rejoin the rest of your network?
It’s the time after VyOS finishes loading, it’s the time from neighbor discovery, exstart, exchange(about 2 seconds), then loading for about 11 min
In this specific area, there’s only about 80 routes for it to load, I’ve removed all the firewall rules, so there’s nothing there causing that. I should mention there’s an OpenBSD router in that same area, and vyos sets up that router and achieves full in about 10 seconds. The two junipers it’s connecting to are only sending defaults as well.
I would probably try to do some packet captures using tcpdump either locally on your VyOS box or better yet setup a portmirror on a switch along the road to figure out whats really going on.
Like is it the VyOS box who takes 11 minutes before sending out the first OSPF frame or is it like the OpenBSD box who takes 11 minutes to reply?
We tried setting all defaults manually just in case, we did find a way to speed it up significantly, however the way is by entering vtysh, then clearing one of the ospf sessions manually, sometimes makes it renegotiate them all much quicker.