Hi to all, I noticed many complains regarding problem of freezing ethernet ports in rb435G. Here is the simple solution which you can use to avoid rebooting and packet loss until MikroTik post fix in ROS:
Go to Switch > switch1 and uncheck: switch all ports
Go to System>Scripts, click on + and, write name: ether
then add this in source:
:if ([/ping 192.168.1.1 interval=1 count=10] =0) do={
log info “ether port is down” ; /interface ethernet switch set 0 switch-all-ports=yes ;
/interface ethernet switch set 0 switch-all-ports=no
}
Please NOTE that 192.168.1.1 should be ip address of any other RB connected to RB435G via Ethernet Port. You can change that according your setup.
Go to System>Netwatch .. click on + and in host field type 192.168.1.1 (ip of other rb via ethernet), interval 00:00:10 and timeout 1000ms … at Down Tab, write: /system script run ether
In Netwatch copy rule at 3. and change interval 00:00:15 (Please note that this step is important as in test I found out that sometimes rb do not respond well to first rule, so you must have this)
You can copy again with interval interval 00:00:20 for any case - this is optional
This is it. I tested this on ROS 5.8 and 435G and it works perfect. You can avoid that watchdoog reboot it so packet loss is much smaller in the moment rb435g ether ports stop responding.
Feel free to test it on 493G but you must use this source:
:if ([/ping 192.168.1.1 interval=1 count=10] =0) do={
log info “ether port is down” ; /interface ethernet switch set 0 switch-all-ports=yes ;
/interface ethernet switch set 0 switch-all-ports=no ; /interface ethernet switch set 1 switch-all-ports=yes ;
/interface ethernet switch set 1 switch-all-ports=no
}
as this model has two chips. Actually I didn’t test this on 493G, but it should work. If you need to use switch-all-ports = yes .. you must change positions of yes and no commands in script.
We’ve recently installed our first Mikrotik RBs - two RB435G with R52Hn as backhaul bridge. Devices configured as bridges, wireless as AP<->station WDS.
Both have this issue, but one of them freezes more often - about one time a day, second freezes approx one time a week.
Did not tried to swap them, weather is pretty cold.
At both sides we have about 10m STP cables, 24HPOW (without POE) for power supply.
Ethernet established 1Gbps, switches by D-Link.
It happens on RB435G but not on RB600A. Might be related to that switch chip in RB435G???
Only happens when ethernet port sends traffic. If it only receives never had happened. Might be ethernet or switch chip Tx counters/buffers overflow ???
When it happens disable/enable the ethernet port doesn’t resolve it. But toggle the state of “switch-all-ports=yes or no” makes the port work again, without a reboot.
Doesn’t matter how port is connected: 10m FTP to a Gigabit dumb switch or 30m FTP to a Gigabit L2 Managed switch both without PoE.
switchA–(etherX-rb435g-r52hn)-> 20km <-(r52hn-rb435g-etherX)–switchB
A–>B 30-180 Mbps (the real traffic)
A<–B 0-20 Kbps (some pings)
..sorry, can’t upload picture.
Traffic is routed one way through that link (between 30 and 180 Mbps).
So one rb435g receives it from switchA at etherX, sends it via wireless to the other rb435g, which then sends it through its etherX to switchB. In the opposite direction through this link goes some negligeable 0-20Kbps(some pings).
The problem is with rb435g whose ethernet sends that 30-180Mbps, the other rb435g whose ethernet sends 0-20Kbps is OK.
When it happens disable/enable the ethernet port doesn’t resolve it. But toggle the state of “switch-all-ports=yes or no” makes the port work again, without a reboot.
you get more data passing over the link, else there is only small amount of traffic passing?
So this affects the rb435g whose ethernet sends out 50-180Mpbs and receives 0-2Kbps,
BUT NOT on other rb435g whose ethernet sends out 0-2Kbps and receives 50-180Mpbs.
Only happens when ethernet port sends traffic. If it only receives never had happened. Might be ethernet or switch chip Tx counters/buffers overflow ???
RB450G has the problem, too. Seems to be related to the Atheros 8316. After 2 minutes of full load (Gbit traffic between 2 ports) the switch stops working.
Did someone try the latest 6.0beta3 ROS on an RB435G/RB450G? Is the switch chip still freezes under heavy load? The changelog indicates that there is a new kernel in that release and drivers were updated.
Thanks exe,
I have the same issue with two Routerboard 435G working as bridges.
The idea is perfect and working.
I hope Mikrotik will fix this issue someday ..
Here I am using four RB435G. All these are configured this way:
one R52Hn card in AP bridge or Station WDS mode
Bridge with all ethernet port, but just with one of them really being used
But they have a difference in configuration: two are using the ether1 port and other two are using the ether2 port
The problem of stop passing traffic was happening with all these RB’s. So, I disabled the “switch-all-ports”.
After this, just the RB’s that are using the ether2 port stop working again.
The RB’s that are using the ether1 port never stop working again.
I put the script described on first topic to monitor the traffic and the “run count” is zero in this case.
So, if you need just one ethernet port, try to use the ether1. Maybe, it can help while this problem is not solved by the Mikrotik.
