Problem with queues

RouterOS General
1

Hello
This is the situation:
Tower Router (CCR 1009)
We have a queue tree system:
Queue parent: TOWER #01
target: IP range /24 of the users

user queue:
IP 1.1.1.1 (included in the range /24 above)
Queue type: DEFAULT
total queue type: DEFAULT
Default type: 512 packets

Interface queue of the eth faced towards customers: multi-queue-eth-default with packet size 512.

Everything works good but I am not able to give sufficent bandwidth to customers.
Everything were OK when i sold 100/100
now with 300 UP-1000 Down, The customers only get about 190mbps. If I disable the queue completely, it gests 800mb.

What is wrong?
ROS is latest long-term

2

No one uses queue?

queue type
set 0 pfifo-limit=1536
set 8 mq-pfifo-limit=512
/queue simple
add comment=TOTAL max-limit=1G/1G name=FO_MZ queue=default/default target=1_FO_Mezzano total-queue=default (FO MEZZANO IS A BRIDGE INTERFACE)
add comment=“sw#02 09/01” max-limit=100M/100M name=“user 01” parent=FO_MZ queue=default/default target=100.70.1.1/32 total-queue=default
add comment=“sw#02 01/12” name=“user 02” parent=FO_MZ queue=default/default target=185.240.69.189/32 total-queue=default
add comment=“sw#02 09/02” max-limit=100M/100M name=“user 03” parent=FO_MZ queue=default/default target=100.70.1.2/32 total-queue=default
add comment=“SWITCH Via Zuccherificio 2” max-limit=100M/100M name=“user 04” parent=FO_MZ queue=default/default target=100.70.1.80/32 total-queue=default
add comment=“SWITCH Via Zuccherificio 2” max-limit=300M/1G name=“user 05” parent=FO_MZ queue=default/default target=100.70.1.84/32 total-queue=default

for example:
the user with 1G download, cannot download more than 180mbit.
If I disable the queue, he goes at 1Gig.
We tried with bandwidth test between the tower router (CCR1009) and the customer devices (HAP AC3). No more than 180m with the queue on, if I disable the queue, it goes to 700m instantly.

3

Really, no one?

4

No none sincerely can help me?

I would like to seek help on queue but I am not getting it rifht and i am not able to find someone to help me.

5

I’d expect the reason to be insufficient CPU for the task, but with CCR10xy, CPU monitoring may be misleading. Nevertheless, the first step is always to run /tool profile, once when you get those 190 Mbps and when you get those 800 Mbps. Also, what exact model of the CCR1009 you’ve got? And how many customers you feed through it?

6

Hello
the issue is randomic.
I have as minimum, che CCR1009
we have maybe 50 to 150 of customers.
I never see one cpu with high utilizations, everyone is at least 15-20 .

I have found in the meantime a partial solution to have all total simple queues, default-small with queue size of 100 and it seem to work even with high speed.