Frequent route calculations and traffice dropouts

RouterOS General
1

Hi,

What causes interface change / route calculations to start?

I am noticing this on some of our P2P Netmetal 5’s. When it happens, traffice is paused for a few seconds. It occurs at random intervals but can happen a many times a minute.

On one run ( two p2p links using nv2 ) I set the STP protocol mode to none and this seems to have solved the problem on that run.

I have noticed it again on another two p2p pair links ( separate network ). The logs on both side show route calculations at the same time.

I have just noticed it happen on a Powerbox router. I don’t know if it cause a pause in traffic though.

If the route calculatiuons are normal, is there anyway to stop the big pauses in traffice?

Cheers
Justin.

This is what the logs show..

May/28/2022 20:28:25 route,debug,event Interface change
May/28/2022 20:28:25 route,debug,event interface=wlan1
May/28/2022 20:28:25 route,debug,event status=UP,RUNNING
May/28/2022 20:28:25 route,debug,event mtu=1500
May/28/2022 20:28:25 route,debug,calc Begin calculation
May/28/2022 20:28:25 route,debug,calc End calculation
May/28/2022 20:30:26 route,debug,event Interface change
May/28/2022 20:30:26 route,debug,event interface=wlan1
May/28/2022 20:30:26 route,debug,event status=UP,RUNNING
May/28/2022 20:30:26 route,debug,event mtu=1500
May/28/2022 20:30:26 route,debug,calc Begin calculation
May/28/2022 20:30:26 route,debug,calc End calculation
May/28/2022 20:30:30 route,debug,event Interface change
May/28/2022 20:30:30 route,debug,event interface=wlan1
May/28/2022 20:30:30 route,debug,event status=UP,RUNNING
May/28/2022 20:30:30 route,debug,event mtu=1500
May/28/2022 20:30:30 route,debug,calc Begin calculation
May/28/2022 20:30:30 route,debug,calc End calculation

2

The event sequence starts with

May/28/2022 20:28:25 route,debug,event > Interface change
May/28/2022 20:28:25 route,debug,event > interface=wlan1
May/28/2022 20:28:25 route,debug,event status=UP,RUNNING
May/28/2022 20:28:25 route,debug,event mtu=1500

which might indicate that there are issues with wlan1 interface state. If this is interface used for PtP link, it might indicate the link is unstable and is toggling active/inactive state transition. The rest seems to be related to BGP, not to xSTP (but I may be wrong on this). So really do verify that wireless PtP link is stable and if it’s not, you need to do something about it.

3

Hi

I have been watching the ptp 2 links and the wifi connection between the two isn’t broken. The log panels in WinBox both scroll at the same time, so if the connection was dropped I would expect the other side to freeze or disconnect from WinBox.

I don’t’ see any log entries that indicate an interface has gone down/up.

I thought it was only happening obn the root side of a PTP link, but it also happens on routers. Usually I can see the interface change log entry for each connected interface.

Could the one change be somewhere in the network triggering or progagating a route recalc over all connected routes devices?


If I had programmed a router using an export of another one and accidentally set a duplicate mac-address somewhere, I assume it would be more obvious. I exported the IP->Neighbours list and couldn’t see any duplicate mac-addresses in it.

I have found a few devices that had their bridge arp mode set to proxy-arp. I set these back to enabled. On one device it kept changing back to proixy-arp.

I am going to turn off stp protocol on a few branches of the network and see if it stabilises, then slowly switch them back on.

4

Some more testing…

I had thought the route calculations may have been causing the traffice pause. With hindsight, that seems ridiculous now.

I am using the bandwidth test (2k bps) from the central router to various routers to see where the pauses are coming from.

I have seen a few route recalculations occur on the remote router being tested without any effect on the bandwidth test.

Also when looking at one router, I noticed the POE-OUT status of some ports of a router was fluttering when there was one of these route recalcs. By ‘flutter’, i mean the status goes blank for a approx 250ms and then comes back. The winbox connection to the device connected to those ports doesn’t drop.