Memory leak in 4.17

Hi,

After MUM in Budapest I’ve upgraded my BGP router from 3.30 to 4.17 - and since that time I was very happy with resoults (for example prefixes where downloaded faster). But yesterday I had a problem with loging to router from bgp. Also from ssh and later mac-telnet.

After few hours I was able to log in for couple of minutes - and router with 1gb of ram mine had only 127 mb free (only routing, 2 full prefix tables).

So - your solution will be to upgrade to 5.x or … ?

M.

Upgrade is always removing of old bugs to make room for new bugs.
Try this, maybe your problem will be resolved.

Enk, but this is not a solution … :frowning:

Did you try it? It is fair game. Either you solve the problem or not. Worth a try.

The issue is not with tring - please notice that when someone has bgp router, this is not a toy to play with, but equipment used to serve services …

I’vd made update to 5.2 - for now, cpu usage is much higher than in 3.30 and 4.17 - with traffic ~20m it has 60%! when on 4.17 it had ~20%.

M.

You right. But this is not related to our discussion. You can make maintance window early in morning for this purpose and notify your users before that. It is de-facto standart.

I’m seeing tha same thing.
I have an RB450G natting 80-100 Mbit/s into 3 vlans.

Since the upgrade from 4.16 to 5.2, the same traffic uses ~20% more CPU.

The funniest thing is that I have high speed PCQ queues, and according to support, this should be much FASTER in v5…



v5 will have higher CPU usage compared to v4 and v3 when router has large routing tables. CPU usage increase depends on how frequent route table update is triggered.

 #      DST-ADDRESS        PREF-SRC        GATEWAY            DISTANCE
 0 A S  0.0.0.0/0                          x.x.65.30          1       
 1 ADC  192.168.1.0/24     192.168.1.1     vlan1              0       
 2 ADC  192.168.20.0/24    192.168.20.1    vlan20             0       
 3 ADC  192.168.21.0/24    192.168.21.1    vlan21             0       
 4 ADC  x.x.65.0/27        x.x.65.27       bridge-wan         0

Is this large? (:


Guys [from mt] you got to be kidding me:

[admin@BGP] > system resource print
                   uptime: 1d21m47s
                  version: "5.2"
              free-memory: 338664KiB
             total-memory: 1024840KiB
                      cpu: "Intel(R)"
                cpu-count: 1
            cpu-frequency: 1595MHz
                 cpu-load: 100%
           free-hdd-space: 86813KiB
          total-hdd-space: 123951KiB
  write-sect-since-reboot: 79138
         write-sect-total: 577054
        architecture-name: "x86"
               board-name: "x86"
                 platform: "MikroTik"

Take a look on free memory and cpu usage!

At this moment aprox 5/5 mbit of traffic on vlans (2 peers - 2 full tables).
And unable to log in from winbox - only ssh and mac-telnet.

On this machine there are no additional filters, mangles, nats and other stuff - just bgp.

And after reboot a supprise:

Login: admin
Password: 
Trying 70:71:BC:9D:F9:C8...
Connected to 70:71:BC:9D:F9:C8
Login failed, incorrect username or password

Connection closed