Hello, I am encountering a weird issue... For whatever reason in Router OS version 6.48.6, only CPU0 is being utilized in RB4011. There are 2 links that are established using PPPoE because both of the modems are bridged. The RB4011 is doing NAT and hosting PPPoE server at the same time. NAT and Queues are being processed only by CPU0, I see this as a problem. There are 3 cores that are left unused. I know that PPPoE is single threaded, but NAT and Queues should have a load being spread all over the cores. I see one similar case posted in the forum as well.

Edit:
We have a HS-concentrator Hex, queues are distributed properly on the cores. It seems that it is only happening on ARM devices?

Edit 2nd:
-Similar cases on ARM devices
viewtopic.php?p=921461 - Trying to debug uneven CPU usage on RB1100AHX4
viewtopic.php?t=172468 - Performance issues with 6.48 and 6.48.1 on ARM/RB4011
viewtopic.php?t=128897 - RB1100Dx4 Multicore issue ? (Though the Router OS version is old, it seems this kind of issue already existed in the past.)

The scope is limited to v6 for now. I can't find any case related to v7 on this matter.

70% load is high, i don't think that neither your PPPoE nor your NAT are causing it...

70% load is high, I don't think that neither your PPPoE nor your NAT are causing it...

There are currently 100+ PPPoE clients that are active in the RB and so is the same for the queues. I'm expecting more throughput on WAN but that's not happening for now because of the core0 being loaded, leaving 1,2, and 3 as idle. I've read that the simple queues are optimized for multi-core but most queues are bound on core0. Is this a bug or maybe ROS is not working properly with ARM devices?

Edit: We have a HS-concentrator that is Hex, queues are distributed through all of it's cores, it seems it is only happening on ARM devices....