Sent to support... fyi

Hello, I have provided autosupouts for 2 rb450s that are hooked together. They are both rebooting with kernal failure every few minutes.

Port 1 on each are hooked together. It may be that the connection between them was 10mb - when I hook them directly together it seems fine. When connected by 10mb link over long fiber connection they reboot constantly?

Thanks,
Scott

How are they being powered? It could be a power supply issue/problem.

You missed the part where when there is a 100mb connetion between the two they work fine. When its 10mb they reboot every couple of minutes (both of them)

Since you asked - using 18v power adaptors plugged directly into the units.

Scott

I have the same problem with RB411's if I set the eth port to 10Mb full or half the board reboots constantly. Anyone have any ideas on this?

Thanks,
Chadd

I have tried version 3.6 and version 3.10 and both have the same problem with the reboot if you try to manually select the ethernet port speed.

In my original post starting this thread I mentioned I already sent all info to support. I have not heard anything back since then so far. I assume they are looking into it.

Scott

also for me,
after 3.10 installation.

[admin@MikroTik] /system routerboard> print
routerboard: yes
model: "133C3"
serial-number: "152701CAE3BB"
current-firmware: "2.15"
upgrade-firmware: "2.15"

jan/01/1970 01:00:21 system,error,critical System rebooted because of kernel failure
jan/01/1970 01:00:19 system,error,critical System rebooted because of kernel failure
jan/01/1970 01:00:19 system,error,critical router rebooted without proper shutdown, probably power outage

a reboot every ~24h.

My RB is powered via POE by a 18 volts 70w max power supply.

please use one of the switch ports (2-5), the last four on the right side.
We will include information in the manual about port 1 and 10Mb/s. We don't have a fix for it as this time.

So the rb450 can be used however I wish (not in bridge or switch mode) so long as the eth1 is not 10mb? I can live with that. 10mb not common anyway, just some special circumstances.

Thanks
Scott

It really makes you wonder what kind of testing is being done on this product before it is released, I mean come on a board with an ethernet port that cant run at 10Mb? It would't be bad if the other boards werent EOL'ed already.

is the 10 Mbps bug confirmed across more boards ?

On my 133C i have reboots also with 100 Mbps eth settings!

But the "bug" isn't -time costant-: some times Kernel Failure happens in 24 h, other times all works fine for 2-3 days.

All started with 3.10 upgrade, that gived me a kernel failure just after installation!

is the 10 Mbps bug confirmed across more boards ?

On my 133C i have reboots also with 100 Mbps eth settings!

But the "bug" isn't -time costant-: some times Kernel Failure happens in 24 h, other times all works fine for 2-3 days.

All started with 3.10 upgrade, that gived me a kernel failure just after installation!

which was the last version where the RouterOS was working ok?
Could you send the support output file after the kernel panic to support@mikrotik.com?

which was the last version where the RouterOS was working ok?
Could you send the support output file after the kernel panic to support@mikrotik.com?

Latest working version was 3.9.

I wait for the next kernel failure and send file to mikrotik.

i've got the same problem with 3.10
my post http://forum.mikrotik.com/viewtopic.php ... 45#p116645

I have a 433 and also see the problem with 3.9 and 3.8.

Eric

I can confirm 433 (w/10mb eth1) had same problem with same fix. Moved the ethernet from the first to third and all is good now. Wondering if that bug affects most of the 3.9 builds.

-eric

I think this bug is RB400 series ether1 specific (Ethernet controller specific). AFAIK this controller is build in into CPU. And it looks like at this point RouterOS don't have support for Half duplex modes. So lets just sit and wait for the next release and driver update for this controller

So the rb450 can be used however I wish (not in bridge or switch mode) so long as the eth1 is not 10mb? I can live with that.

I did a new install last week, two RB433 and an RB333. All three on ROS 3.10. All three at 10MB on port 1.

The RB333 was fine, both of the RB433s exhibited the symptoms described on this thread. Bumped them both up to 100MB on port 1, and they behaved much better, but even then had a couple of tests where the ports became unresponsive.

At 10MB, I could easily force the ports to become unresponsive just by running the bandwidth test at speedtest.net. The download test usually worked fine, but port 1 would almost always become unresponsive during the upload test.

While using one of the other ports may get around this problem, it was not an acceptable solution to me because I was powering the routerboards using PoE, which is only available on port 1.

So I really hope this gets resolved soon!

It should now be fixed with RouterOS 3.11 and the 2.16 RouterBoot. (Be sure to upgrade the firmware.)

What's new in 3.11:

*) fixed bug - in some cases web proxy https with parent-proxy did not work;
*) added default-route-distance setting for DHCP client;
*) mesh protocol - bridge interface in a mesh did not work well, fixed;
*) multicast - fixed bootstrap router (BSR) mechanism;
*) user manager - users can now be redirected to HotSpot login page after
PayPal payment;
*) added ability to dst. nat only address or port, not both at the same time;
*) ospf - fixed default route;
*) ipsec - fixed tunnel mode with dynamically generated policy;
*) port remote-access - fixed allowed-addresses check;
*) ethernet half duplex modes on rb400 series work now;
<cut>

Tom