MikroTik

Being full working network interface simply ceases to function e not responding. I have to reboot the to the same return, 10 equipment with the problem. Versão 5.7 - 2.36.

[Ticket#20​1112026600​0099]

My observations show that this thing happens ONLY when ethernet interface sends allot of traffic(80-150Mbps). If that same ethernet ONLY receives that same traffic there is no problem.
Like this:
siteA--(RB435G_A)----wireless----(RB435G_B)--siteB
one way real world traffic (the other way is through another link with RB600A and have never developped that problem):
if siteA sends to sideB only RB435G_B have that problem 2 to 5 times per day;
if siteB sends to sideA only RB435G_A have that problem 2 to 5 times per day;

Reboot is one way to solve it , the other is to toggle the state of "switch-all-ports" to yes/no (in interface ethernet switch)

So I suppose it has something to do with some TX counters overflow and something with that switch chip in RB435G which is not present in RB600A.

Hope this gives someone a clue

same at ticket [Ticket#2011120166000279]

what about v5.10 ? is there fixed this problem? thanks

2011120166000279 is unrelated to the issue. it talks about nv2, not about RB435G
20​1112026600​0099 there is no such ticket

from my ticket request ([Ticket#2011120166000279]):

> It's nice to see solid nv2 on ROS5.9 working, but all over the wireless
> pleasure and speeds i'm still having issues with ethernet interfaces,
> which stop
> possibilities of nv2 to work with it. As same as topic on
> http://forum.mikrotik.com/viewtopic.php ... 50#p285372,
>
> i have 1g ethernet interfaces freezing(actually stopping to transmit data)
> same on 5.5 5.6 5.7 5.8 5.9 of ROS, but have a solid ethernet connections
> on 4.17.
ITS NOT about NV2, its about ethernet freezing

I'm having same/similar problems.

Only a reboot fixes the problem. Also noticed is that when counters are cleared on the interface even though there's TX packets going No RX packets. The counters do not increase..

Any idea on the resolution of this Mikrotik ? Love the product this right now is beating us up. We have over 11 PTP links running 5.11 all which have experienced the loss of connectivity on the ethernet side of the remote radio.

I was the MUM-SP and talked to many people. It has the same problem as others. The MikroTik delay to talk about the problem. Support MikroTik ta ta too weak.

2011120166000279 is unrelated to the issue. it talks about nv2, not about RB435G
20​1112026600​0099 there is no such ticket

http://forum.mikrotik.com/viewtopic.php?f=1&t=57215

We'll have to use the wireless card and network card? As in this photo?

I'm having same/similar problems.

Only a reboot fixes the problem. Also noticed is that when counters are cleared on the interface even though there's TX packets going No RX packets. The counters do not increase..

Any idea on the resolution of this Mikrotik ? Love the product this right now is beating us up. We have over 11 PTP links running 5.11 all which have experienced the loss of connectivity on the ethernet side of the remote radio.

You had any response from Mikrotik?


Because I did not.

My observations show that this thing happens ONLY when ethernet interface sends allot of traffic(80-150Mbps). If that same ethernet ONLY receives that same traffic there is no problem.
Like this:
siteA--(RB435G_A)----wireless----(RB435G_B)--siteB
one way real world traffic (the other way is through another link with RB600A and have never developped that problem):
if siteA sends to sideB only RB435G_B have that problem 2 to 5 times per day;
if siteB sends to sideA only RB435G_A have that problem 2 to 5 times per day;

Reboot is one way to solve it , the other is to toggle the state of "switch-all-ports" to yes/no (in interface ethernet switch)

So I suppose it has something to do with some TX counters overflow and something with that switch chip in RB435G which is not present in RB600A.

Hope this gives someone a clue

Hi to all, i have same problem on four RB435G!!!! - M I K R O T I K - W A K E - U P!!!

The problem I think is now clear!

You had any response from Mikrotik?

he didn't contact mikrotik

Hi to all, i have same problem on four RB435G!!!! - M I K R O T I K - W A K E U P!!!

please tell me your support ticket number and I will check the issue. we can't fix something, if you don't give us any information.

You had any response from Mikrotik?

he didn't contact mikrotik

Hi to all, i have same problem on four RB435G!!!! - M I K R O T I K - W A K E U P!!!

please tell me your support ticket number and I will check the issue. we can't fix something, if you don't give us any information.

I am returning the 10 435g. And I'm buying the RB800 because they do not have the same problem. Possibly it may be a lot that came with this defect.

we are having the same ethernet freze on rb435g.

When the eth frezzes we can access the rb by wireless interface and the freezed eth is reported connected but no traffic goes through.

Enabling conncetion tracking seems to reduce the freeze issue from some hours to some days but it is not a real fix.

regards
ros

I can only say one fing send in supout files to mikrotik so they have a chans to fix the problem.

I openede the following ticket [Ticket#2012022166000249] including supout taken after the eth port freezed and flapped

regards
Ros

I can only say one fing send in supout files to mikrotik so they have a chans to fix the problem.

we don't have any emails from you to support. please paste ticket number

I openede the following ticket [Ticket#2012022166000249] including supout taken after the eth port freezed and flapped

Thank you for the report.
We are working on the problem.

please as soon as you have a test version to fix this issue let me know to test it.

regards
Ros

Hi,
I have the same problem with my 435g, same days frezzes 1 of them and another day the other... it's products in delivering internet to real people and it's not some testings...
Is there a way to solve this issue? Or must i change to another product?
Current version is 5.12 and the problem is happening day after day.
Best Regards,
Nuno

Hi,
I have the same problem with my 435g, same days frezzes 1 of them and another day the other... it's products in delivering internet to real people and it's not some testings...
Is there a way to solve this issue? Or must i change to another product?
Current version is 5.12 and the problem is happening day after day.
Best Regards,
Nuno

contact support. this is user forum.

Hi,
I have the same problem with my 435g, same days frezzes 1 of them and another day the other... it's products in delivering internet to real people and it's not some testings...
Is there a way to solve this issue? Or must i change to another product?
Current version is 5.12 and the problem is happening day after day.
Best Regards,
Nuno

enable connection tracking to have the issue happening once a week, not once a day.

waiting about the final fix from MT.

regards
Ros

Thanks rpingar, that what i'm start doing now!
Best Regards,
Nuno

Thanks rpingar, that what i'm start doing now!
Best Regards,
Nuno

if it helps you, please give us a karma.

regards
Ros

the 5.14 release seems to fix the flap issue.
Now 5 hours of running without issue.

regards
ros

it crashed after 7 hours

So no progress about it.

Using 5.13 the gbit eth of the rb435, using the MT gbit poe and 13v power, flaps once each two days.

Using 5.14 the gbit in the same identical condition freeze, causing a lot of troubles.
Someone wrote that 5.14 fixed the flap issue on rb750g so i think there is something new about the eths. But it doesn't fix the rb435g eth issue.

Any update on the resolution of the ticket: [Ticket#2012022166000249] ??????

Thanks
Ros

we probably find the cause of port flap/freezing on rb435g.

It seems to be ralted to snmp high usage or the cpu usage snmp is causing on havy traffic load.

Infact we noted the the big part of the port flap issue were reported during snmp polling.
So yesterday evening we disable the snmp and also disabled the connection tracking, to let the issue happen, more frequently.

Since we disabled snmp we didn’t get any port flap or freeze. They are only 12 hours but anught to let you know our experience.

hope this can help to find and fix the issue.

regards
Ros

Hello!
Is there at least some solution to this problem?
We have a large party rb435. It is not nice: (
Power POE (22 v), and the second ethernet port

I have same problem with all routerboards with gigabit ethernet ports (RB711G-5HnD, RB435G, RB433GL etc.). I tryed more of firmwares. The firmware 5.5 is the most "stable".

we removed the 435 and usend only rb800

MT this time you missed the point!

regards
Ros

We just bought our first pair of 435G boards and they're both doing this! Both of them are powered and connected via the MT gigE PoE injector. One of them plugs into an RB250GS, but the other into a Soekris router. The switch-all-ports toggle does bring it back to life as reported elsewhere, but obviously this is a crazy fix.

I guess we'll be RMA'ing until we get a stable one, or MikroTik give us a solid response about this problem.

Please contact support if you have any RB435G ethernet issues. Include a supout.rif file from RouterOS v5.14

normis you should have around 15 tickets opened about it since 2 months ago.

I think it is time to put pressur on the developers to fix this issue.

regards
Ros

no, we don't ! that's why I keep asking for those tickets.
currently we have not been able to repeat the issue even following exact steps.

it is mine:
[Ticket#2012022166000249] rb435g gbit eth problem

with supout and comments.

Regards
Ros

Yes, I know. You are one of 2-3 people who have contacted support about this.

[Ticket#2012032866000387]

We're going to be purchasing another board for lab testing. If we find a method of easily reproducing the problem in a lab environment, I'll post it.

Any one of you use the Dude to monitor these units ? Or anything else that uses SNMP?

yes we use snmp, and disabling snmp seems alleviate the problem without fixing completely it.

regards
Ros

We use SNMP too.

We are doing some more tests. Let us know some other details, which SNMP version you use, if you monitor specific OID, how many wireless cards do you have active etc.

we have 1 wireless card and monitor the follwing parameter:
- cpu usage
- hd (used/total)
- memory (used/total)
- registered wireless client
- data rate
- wireless signal
- frequency

regards
Ros

1 wireless card and we monitor only wireless signal strength through snmp.

We use SNMPv1.

OIDs we poll are:

IF-MIB::ifInOctets.X
IF-MIB::ifInUcastPkts.X
IF-MIB::ifInNUcastPkts.X
IF-MIB::ifOutOctets.X
IF-MIB::ifOutUcastPkts.X
IF-MIB::ifOutNUcastPkts.X
1.3.6.1.4.1.14988.1.1.1.1.1.4.X
1.3.6.1.4.1.14988.1.1.1.2.1.3.Y.X
1.3.6.1.4.1.14988.1.1.1.2.1.4.Y.X
1.3.6.1.4.1.14988.1.1.1.2.1.5.Y.X
1.3.6.1.4.1.14988.1.1.1.2.1.6.Y.X
1.3.6.1.4.1.14988.1.1.1.2.1.7.Y.X
1.3.6.1.4.1.14988.1.1.1.2.1.8.Y.X
1.3.6.1.4.1.14988.1.1.1.2.1.9.Y.X
hrProcessorLoad.1
enterprises.14988.1.1.3.8.0
enterprises.14988.1.1.3.10.0
hrStorageDescr (walked)
hrStorageSize.Z
hrStorageUsed.Z

In above, X is the interface ID - either an ethernet interface, or a wlan interface, or both in some cases.
Y is a station MAC address.
Z is a storage ID - only the main memory ID which we discover from walking hrStorageDescr.

Some of the above OIDs are net-snmp specific abbreviations. Hope that's OK... I'll resolve them to numeric OIDs if you need.

And in both 435Gs we have only one wlan card installed and active - Ubiquiti SR71-15.

any update?????

regards
ros

any update?????

Not sure about you, but I enabled a netwatch script to toggle the switch-all-ports option if the ethernet interface died, and just having the netwatch running has prevented the interface from going dead on both my 435Gs (script hasn't fired once). Go figure. If it remains stable for a few more days, I'm going to try disable the netwatch to see what happens.

still no update

We also bought a large number of RB435g, and almost all the time 3-7 days stuck and have to constantly restart!
If you hang the device only works WAN does not work LAN,
disabled almost all the functions SNMP, Tracking,
Each has a 435 to 10 client devices!

I'm having same problem with my boards locking up .. I have watchdog running and set to watch the main gateway .. it will auto-reboot, but I can't get it to do supout.rif so that I can send in the supout for tech help

James

All I can say is, those who are still experiencing it, please help MT debug it with remote access to your equipment. Our boards haven't frozen for over a week now, and we haven't really changed much. As a result, they feel like ticking time bombs, but the problem is impossible to debug like this...

Hi,
we had report this issue to MT support in December: Ticket#2011120566000486

We had two RB435G used for our main backbone (200 Mbps link). The link stop to work almost every days, as it's been always more than 50 Mbps going throught it ... The only way to get the link back to normal state is to disable/enable switch-all-port feature or by reboot the RB435G.

I had granted access to their developper recently (last week) but still got no news. I think it could help if at least someone from MT could tell us:

1. If they found the problem
2. If it depend on linux kernel or driver, or is it a routeros issue.
3. If they got an idea of how to fix it and how many time it will take to fix it (day,week, month, year .

We suffer much by waiting without any news, and let me say you that our customer suffer too... If the problem is too hard too resolve quickly, please tell us and we will take action to resolve it by replacing these board by another without switch chip (RB711GA-5HnD).

Regards,
Michael

Hi,
we had report this issue to MT support in December: Ticket#2011120566000486

We had two RB435G used for our main backbone (200 Mbps link). The link stop to work almost every days, as it's been always more than 50 Mbps going throught it ... The only way to get the link back to normal state is to disable/enable switch-all-port feature or by reboot the RB435G.

I had granted access to their developper recently (last week) but still got no news. I think it could help if at least someone from MT could tell us:

1. If they found the problem
2. If it depend on linux kernel or driver, or is it a routeros issue.
3. If they got an idea of how to fix it and how many time it will take to fix it (day,week, month, year .

We suffer much by waiting without any news, and let me say you that our customer suffer too... If the problem is too hard too resolve quickly, please tell us and we will take action to resolve it by replacing these board by another without switch chip (RB711GA-5HnD).

Regards,
Michael

+
And i have the same problem.

Running 5.15 for 2 weeks and that ethernet lockup is still present. About once a day.

And I doubt 5.16 will make any change.

I too have some outages of ethernet in rb1200. Router itself remains up but I cannot access it by any means for a while. Unfortunately while I'm going to investigate with a laptop it resumes normal operation. Any ideas how to check what's wrong?

Running 5.15 for 2 weeks and that ethernet lockup is still present. About once a day.

And I doubt 5.16 will make any change.

So, 5.17 and rb435g and ethernet lockup is still there.

Hi,
just to let you know I solved my problem.

Here is our setup:

site1 <---> RB493AH <---> RB435G <---> over wireless NV2 <---> RB435G <---> RB450G <---> site2

On RB493AH (ROS 5.7 and FW 2.20) and RB450G (ROS 5.14 FW 2.39) all port are in a bridge interface.
On both RB435G (ROS 5.14 FW 2.39) wlan1 and ether1 are in a bridge interface.
No other special setup.
As describe previously, after a random time, data stop to passthrough ethernet between RB435G and RB450G and some time between RB493AH and RB435G. As describe previously, problem seems to be an issue with the switch chip as playing with the switch-all-port parameter let the traffic go through the ethernet port again.

To fix this issue, if you don't need to ethernet switch-all-port:
- put the setting to no;
- pay attention to the ethernet synchronisation as sometime it fallback to half duplex;
- if so, force ethernet speed/duplex re-negotiation;
- enjoy the uptime of your ethernet link...

No more ethernet flap since may 29 2012, but I had to mention that the port on site 2 had negotiate 100 Mbps FD instead of 1Gbps FD ?! We will verify a second time lightning (gigabit one), POE (gigabit one) and CAT6 cable to be sure and let you know about all developpement.

Regards,

Michael Plourde
Digicom

P.S. Don't forget to give karma

I too have some outages of ethernet in rb1200. Router itself remains up but I cannot access it by any means for a while. Unfortunately while I'm going to investigate with a laptop it resumes normal operation. Any ideas how to check what's wrong?

Probably this not related to the same issue.
1.What is the purpose of this router ?
2.Do it have public IP ?
3.Any firewall ?
4.All ethernet ports going down at the same time ?

- pay attention to the ethernet synchronisation as sometime it fallback to half duplex;

Did you have something mentioned about half duplex in the log, because I have never seen a line stating that in mine.
It is always "interface,info ether3 link up (speed 1000M, full duplex)"

- pay attention to the ethernet synchronisation as sometime it fallback to half duplex;

Did you have something mentioned about half duplex in the log, because I have never seen a line stating that in mine.
It is always "interface,info ether3 link up (speed 1000M, full duplex)"

If you don't then everything is good. sometime it fallback to half duplex when you remove the switch-all-port option. But if not then everything is fine, enjoy !

For me, it's now 20 days without suffering port flapping...

@digicomtech
your method (switch-all-port=no) doesn't work for me, I'm sure the ethernet is not on half duplex and it still freezes 1-3 times per day. I have tested it couple of days with no luck .

So I'm interested if this works for someone else with rb435g ?

@digicomtech
your method (switch-all-port=no) doesn't work for me, I'm sure the ethernet is not on half duplex and it still freezes 1-3 times per day. I have tested it couple of days with no luck .

So I'm interested if this works for someone else with rb435g ?

Setting switch-all-port=no works on about 90% of the 60+ RB435G boards I have in my back haul network. We have noticed that it is only an issue if 10mbps+ bandwidth is being pushed through. The boards that this setting did not fix have been switched to RB800 boards, with which we have never had an issue.

My observations show that this thing happens ONLY when ethernet interface sends allot of traffic(80-150Mbps). If that same ethernet ONLY receives that same traffic there is no problem.
Like this:
siteA--(RB435G_A)----wireless----(RB435G_B)--siteB
one way real world traffic (the other way is through another link with RB600A and have never developped that problem):
if siteA sends to sideB only RB435G_B have that problem 2 to 5 times per day;
if siteB sends to sideA only RB435G_A have that problem 2 to 5 times per day;

Reboot is one way to solve it , the other is to toggle the state of "switch-all-ports" to yes/no (in interface ethernet switch)

So I suppose it has something to do with some TX counters overflow and something with that switch chip in RB435G which is not present in RB600A.

Hope this gives someone a clue

I think you're right. We suffer from a DDOS attack on a client at the remote end of our bridge setup and the ethernet stop to pass the data when packet per second reach a too high value.
Maybe, you should graph you packet per second on the RB435G ethernet port.

Regards,
Michael

@digicomtech
your method (switch-all-port=no) doesn't work for me, I'm sure the ethernet is not on half duplex and it still freezes 1-3 times per day. I have tested it couple of days with no luck .

So I'm interested if this works for someone else with rb435g ?

Setting switch-all-port=no works on about 90% of the 60+ RB435G boards I have in my back haul network. We have noticed that it is only an issue if 10mbps+ bandwidth is being pushed through. The boards that this setting did not fix have been switched to RB800 boards, with which we have never had an issue.

If you watch the spec of both board you will notice that the maximum packet per second that can hold a RB800 is significately higher than an RB435G. Maybe, you got a lot of small packet that the RB435G failed to handle...

Regards,
Michael