Approximately three weeks ago one of the Ethernet interfaces on our RouterOS i386 PC stopped responding. Since then it keeps on stopping randomly, approximately once a day sometimes up to three times a day. The only remedy is to reboot the PC, just disabling and re-enabling the interface or changing the switch port does not help.
We had fairly old PC hardware, 2 x RB44 cards, and RouterOS 2.9.51. The PC is a gateway to various ether segments on our network and also a transparent proxy server. We also experienced fair amounts of ping loss through this PC.
Thinking that we’ll eliminate the router, we upgraded the hardware, added 4 x RB44G cards, and loaded RouterOS 3.11. Thank goodness the ping loss stopped but unfortunately the interface (called Public) still stops responding. We then replaced the Netgear managed switch to which the Public interface was connected. This still didn’t make any difference.
Suddenly, this weekend, the symptoms changed - instead of having the Public interface that stops responding the PC rebooted every two to three hours. We have a syslog running and nothing is logged.
Then Monday, the reboots stopped and we’re back to the original problem. In my experience Linux and RouterOS doesn’t reboot very easily, unless you have some major hardware issue. But we changed all the hardware! We’re thinking of getting a specialized cable tester to come and test every single cable.
I had the same problem when i had a D-Link 530T+ in my box.
It would stop responding and i would have to reboot it.
Now I have a RB44 and it works flawlessly.
The public interface is on one of the RB44Gs. I have tried various ports on the RB44G but I’ll try other ports again.
There are quite a lot of rules for this interface, especially firewall accept and SRC-NAT rules.
Yes the interface gets quite heavy traffic, it’s next gateway is a fiber optic link and 4 ADSL load balanced routers.
I have been contemplating firewall rules or configuration issue, but since it ran before for many years and since it happens so intermittently (today 3 times, sometimes only once in two days), I’m trying to avoid reconfiguring the whole router from scratch.
Hilton, you might be onto something. When I disabled auto-negotiation and switched to 100mbit the symptoms that I have been experiencing immediately appear - data on the Tx side but nothing on the Rx side. (it’s a gigabit interface). I’m going to play around with this.
The funny thing is it had version 2.9.51 and upgrading to 3.11 with completely new motherboard, RB44G card etc still same problem. But I see version 3.12 is out now maybe this will help as well.
For now I’m hoping Hilton could be onto something. Thanks everyone for replying, the community support rocks!
I would disable whatever rules you can and see how it fairs, I know with my old D-Link card when i started moving a lot of traffic through it with all my src-nat rules and filter/mangle rules it would lock up. But that was fixed by the RB44 i got. And I would assume that the RB44G would have done just as well, but been GigE.
Got another box you can test with?
I would copy all the rules and settings to another box running almost the same hardware and software and hammer it with traffic till it failed.