bandwidth-test on RB450G

RouterOS General
1

I have a strange issue here…

One of my clients has a RB450G running ROS 3.25 configured at it’s office as a simple bridge (we use MT’s for link monitoring) and at our end we have a x86 running RouterOS 3.24… There is simple routing between them achieved by a Cisco Catalyst Router.
They are connected together trough an 10Mbps VDSL link witch has currently no profile (its synced at 96/40)

When I do bandwidth-tests from the RB450G to the x86 strange stuff happens:

I know for sure that the x86 end is working fine since it’s reporting correct results for any other links:

See by yourself:

Testing fixed speeds:

[admin@RB450G] > tool bandwidth-test 208.85.113.13 direction=both remote-tx-speed=20M local-tx-speed=20M duration=30s
                status: done testing
              duration: 30s
            tx-current: 10.1Mbps
  tx-10-second-average: 10.1Mbps
      tx-total-average: 10.0Mbps
            rx-current: 20.0Mbps
  rx-10-second-average: 20.0Mbps
      rx-total-average: 20.0Mbps
          lost-packets: 0
           random-data: no
             direction: both
               tx-size: 1500
               rx-size: 1500

Removing fixed speeds (strange - tx speed is lower) :

[admin@RB450G] > tool bandwidth-test 208.85.113.13 direction=both duration=30s                                       
                status: done testing
              duration: 30s
            tx-current: 4.7Mbps
  tx-10-second-average: 4.7Mbps
      tx-total-average: 5.0Mbps
            rx-current: 91.6Mbps
  rx-10-second-average: 90.7Mbps
      rx-total-average: 65.3Mbps
          lost-packets: 829
           random-data: no
             direction: both
               tx-size: 1500
               rx-size: 1500

Setting only remote-tx-speed (seems like the other end is not cooperative with this limit - again strange tx results):

[admin@RB450G] > tool bandwidth-test 208.85.113.13 direction=both remote-tx-speed=20M duration=30s                                       
                status: done testing
              duration: 30s
            tx-current: 4.8Mbps
  tx-10-second-average: 4.7Mbps
      tx-total-average: 4.9Mbps
            rx-current: 82.8Mbps
  rx-10-second-average: 81.1Mbps
      rx-total-average: 59.5Mbps
          lost-packets: 796
           random-data: no
             direction: both
               tx-size: 1500
               rx-size: 1500

Setting local-tx-speed (oh well… stranger than ever):

[admin@RB450G] > tool bandwidth-test 208.85.113.13 direction=both local-tx-speed=20M duration=30s
                status: done testing
              duration: 30s
            tx-current: 10.1Mbps
  tx-10-second-average: 10.2Mbps
      tx-total-average: 10.1Mbps
            rx-current: 19.9Mbps
  rx-10-second-average: 19.9Mbps
      rx-total-average: 20.0Mbps
          lost-packets: 0
           random-data: no
             direction: both
               tx-size: 1500
               rx-size: 1500

Forcing tx speed to a lower value also reduces the throughput ? WTF!?:

[admin@Technicost] > tool bandwidth-test 208.85.113.13 direction=both local-tx-speed=10M duration=30s 
                status: done testing
              duration: 30s
            tx-current: 6.4Mbps
  tx-10-second-average: 6.4Mbps
      tx-total-average: 6.4Mbps
            rx-current: 9.9Mbps
  rx-10-second-average: 9.9Mbps
      rx-total-average: 10.0Mbps
          lost-packets: 0
           random-data: no
             direction: both
               tx-size: 1500
               rx-size: 1500

Why I am having such strange results ? Any clues ?

2

interested!!!

“Forcing tx speed to a lower value also reduces the throughput ? WTF!?:” I have a similar problem!!!

3

Odd…

By the way… Watch out running bandwidth test too much / too long…

Some ISPs (Cox) will “see” the traffic and then start poking arround.. they may see a device they dont recognise (MikroTik) and SHUT YOU OFF !!!

Been ther , Done that !!! last week…

So if you connection DIES after diagnosing a link, Hello support ??? can you turn it back on…

4

Bandwidth-test tries to use all bandwidth available. Running UPD BTest both direction without limitation could cause packet drops and that means there will not be complete synchronization between BTest client and server. Do you experience any problems running TCP BTest or single direction UPD BTest?