RB1100Dx4 Multicore issue ?

RouterOS General
1

Hi,

I got a brand new RB1100Dh4 a few days ago, I`ve noticed that shes running with only one core at 40~80% while the others are at 0~1% usage

i had a RB750G3 before and using all 4 cores doing the same job.

the RPS feature is enabled on all interfaces “came disabled by default”
cpu usage2.png

2

Depends on the tasks and processes running. Make more queues or other things, it will start to use other cores.

3

It’s all so)
IMHO If they do the load distribution - there will be problems with reordering the packets.
“ARM” architecture is not yet working normally.

4

You should try 6.41 (http://forum.mikrotik.com/t/v6-41-current/114978/1)

What’s new in 6.41 (2017-Dec-22 11:55):
*) arm - minor improvements on CPU load distribution for RB1100 series devices

Might solve this issue for you.

5

Hey Normis,
They work! ive seen with dude running. i was just hoping for more, i mean in a few months ill have to upgrade again with my routerboard running with 90% usage on core “0” while the others at 1%
well should i aim for a x86 ?

Thanks for your reply

6

Hey Kindis

I was so excited when i read it, but didnt work

Thanks for your reply tho
update.png

7

Show us /tool profile

This way we can see what is using the CPU core.

8

Sure
cpu profile.png

9

This is weird. I was hopping to find one process eating almost all CPU. Instead I found 4 or 5 processes, using only one core.

  1. Your firewall is eating 30% of a single core. I believe the Mikrotik implementation spread the load assigning one stream to each core. Are these numbers the result of a single stream? One single connection, downloading something? If yes, that would explain it. If this is the case, try with a torrent - they open dozens of connections, and the load should use all 4 cores.
  2. What is the connection speed? 30% of a single core is quite something. Maybe some optimization on the firewall/queue rules? I know this isn’t the issue, but…
  3. You may be right, and this is a bug. I don’t have one of this units to test, so can’t help this way.
10

yeah, but the firewall rules are simple, like routing mark, drop bogons, and incoming ssh telnet ftp from wan, some nat redirect. nothing special

I`ve 3 wan with 300Mbps each
the router show more than 18.000 connections atm (about 400 customers)

11

This is weird. To me, either is a bug, or something with the queues - but I have no experience with them, so can’t help with this. :frowning:

12

Should be like this.. right ?
RB HEX.png

13

Yes, something like this.

14

Well as predicted the SINGLE CORE had achiev 100% usage

i’m so dissapointed with this routerboard

i’ve 15 rules on firewall filter only.
drop bogons list
drop invalid conn
drop telnet-ssh-dns from wan

on nat tab
masquerade wan 1, wan 2, wan 3
some dst nat

mangle tab
only routing mark for wan2 and wan3
NICE100%.jpg

15

From Firewall / Connections / Tracking set TCP Established Timeout to something small, like 5 minutes to at least drop connections that aren’t doing anything. This will not effect active connections. Does this help?

16

Yes it helped, the core is running around 94%. still only core 1 running tho.

a brand new ccr1009-7g-1c-1s+ its on the way. i hope this one will solve my problem.

17

I just want you guys to know the CCR1009 just solved the problem. every single core works!
awesome2.png
awesome.png

18

Good to hear.

19

Did you log a ticket with support? I am sure they would be interested to know.