CCR2216 BGP Full-Table CPU load not behaving deterministic

We have bought a CCR2216 to test its potential as a BGP router.

Right now we have set up a lab where we receive a full table from our peer and send out just 2 /24 networks ourselves.

90% of the time when we start up the BGP session, it works perfectly with the CCR2216 showing the full table in under 30 seconds in its routing table.

The other 10% of the time the router takes up to 5 minutes to show the full table in its routing table.

When comparing the two scenarios we see that 10% of the time the router does not seem to distribute the load between all CPUs evenly, rather using only one of the CPUs for the routing process.

Interestingly we have seen it even switch the used CPU while the session was still active, although it is not always using the same CPU when this occurs.

After rebooting the CCR2216 and starting the BGP session again the load is then distributed between the CPUs until we disable and enable the session a few times, obviously with some minutes in between until the router has stopped showing high CPU load from building / deleting its routing table and the problem occurring again.
Config.txt (2.03 KB)

Try setting input.affinity and output.affinity to alone instead of main.

Then it should work better..

I tried that but unfortunately the problem can still be reproduced, thank you though.

I faced a similar problem with the 2116 router, where the CPU occasionally spikes to 100% for 2-3 minutes while I run eBGP. Have you found a solution yet?

same issue, i opened a ticket
has someone found a solution?