Router Crash Serveral times a Week

jmedinas · Mon May 09, 2005 6:07 pm

Hi guys,

I have a problem that started 2 weeks ago.
The router was working very well with version 2.8.20 and one day it crashed.
In the monitor shows a Kernel Panic with problem in the "swapper" process.
The lights was blinking in the keyboard.

I reboot and 2 days after the same problem, rebooted again and 3 hours after the same problem and it begins everyday it crash.

I switched the machine complety for another new an the problem continues
I upgrade the version to 2.8.27 in both machines and the problem continues
I switch ethernet cards , memory , disk , mikrotik licenses and the problem continues

What can it be ? Where I can find more information about the crash in the router... Any tips ?

This is only a Ethernet Router... with 3com and intel cards (5)
the cpu load always is 14 - 20

thank you!

changeip · Mon May 09, 2005 6:45 pm

You did physically swap RAM for another set of RAM? Run memtest86 on that machine to see if there are any RAM errors. You will be surprised at how many people have bad ram and don't know it, especially after 1-2 years in use.

Sam

jmedinas · Mon May 09, 2005 7:46 pm

Yes I swap the RAM with another and the router crash again.

The original ram I put it in another machine and RUN the memtest86 that come with the Debian Cd i let the test for 20 hours and no error was reported.

I change algo all the cards , the HDD , the case , the MoBo , everything ...

I suspect that is software bug or attack ?

any others ideas please....

zack · Mon May 09, 2005 8:07 pm

Strange.
I have the same problem. And the problem started 6-7 days ago. Before everythind was just fine. Then my Wireless routers 2.8.21, 2.8.22 started to restart themselfs and somethimes to crash. Can somebody help us.
My config is only with wireless cards and no Ehternet.

zack · Mon May 09, 2005 9:14 pm

Is there any way the router to restart itself when it crashes

jmedinas · Mon May 09, 2005 9:26 pm

You need to activate the Watchdog feauture , but I think it only works with RouterBoard from mikrotik.

/system watchdog

My router with the crash problema is Intel MoBo and I try with another too with the same results.

Cameron Earnshaw · Tue May 10, 2005 9:12 am

I've been having this problem for weeks now with V 2.8.25 and find the problem goes away when I turn off pcq. I too have tried changing machines. I am now looking for another way to do traffic shaping. I am beginning to think Mikrotik O/S has become unstable but this is not a popular opinion to have on this list it seems.

zack · Tue May 10, 2005 12:59 pm

Does anybody have any progress about the problem.

mp3turbo2 · Tue May 10, 2005 2:56 pm

solution? progress? jmedinas (who started this thread) didn't even send supout file to mikrotik, what kind of support can we users provide when even author of software was not asked to? :)

> This is only a Ethernet Router... with 3com and intel cards (5)

this is what got my attraction. How the hell are you going to stack up 6 ethernet cards in one machine, meaning at least 5 of them on PCI bus, when you have maximum four hardware interrupts on PCI bus? There has to be a lot of IRQ sharing. Also, I'd like to see board with 5 or 6 bus-mastering PCI slots...

Please post your _detailed_ HW configuration. Do not omit EXACT type of motherboard and BIOS you use (ie Abit BH-6 ver.2, Intel 440BX chipset, BIOS version 2.0 codename SX dated 09/26/2002). Please post your EXACT BIOS settings also, try to take photos of screen and put them somewhere on Internet so I can download. I will then advice you what to change.

Also, due to interrupt sharing, I highly recommend you to either

a) split this box into two boxes with 3 cards in each
b) lower amount of cards used in one box, six is too much. You don't have to have 6 physical cards when you have 6 logical subnets.

I have ONE ethernet card with 20 TCP subnets on it. Clients do not see each other and everything is fine.

bye, mp3turbo.

tully · Tue May 10, 2005 3:18 pm

Most of the newer 3com cards actually have the Realtek chip (I think). We have always suggested not to use Realtek Eth chips. Change these cards and the problem will probably be gone.

John

Cameron Earnshaw · Tue May 10, 2005 6:41 pm

Tulley, what is your preferred NIC? Ours seem to be Accton/SMC. Also, what about the integrated NIC on the PC motherboard? Should we avoid using it? I would be a happy man if this problem could be resolved by changing NICs.

changeip · Tue May 10, 2005 6:49 pm

We always prefer Intel NICs, server version if possible. 3COM went downhill and is no longer any good, realtek sucks, and anything else just doesn't compare.

Still testing broadcom and marvell nics.

Sam

jmedinas · Wed May 11, 2005 7:52 pm

solution? progress? jmedinas (who started this thread) didn't even send supout file to mikrotik, what kind of support can we users provide when even author of software was not asked to? :)

I send the supout file to mikrotik and I start a TT [Ticket#: 2005050916000285] before I post here.... I still have the problem. The router reboots itself 3 times yesterday and came back alone with the Watchdog feature but today in the morning doesn't come online again.

The response from Mikrotik was to take a picture from the screen with the kernel panic. I hope the next time it crash I have the camara and take the time to picture it and reboot -the Router is about 1 hour from my office-.

I will follow your idea to split the box. I have physical 6 ethernet cables to backbones (3) aps (3) that I wish to leave it alone.

Thanks !

jmedinas · Wed May 11, 2005 8:10 pm

Specs.

Router A (The Spare one, currently online)
Flash IDE from Mikrotik
RAM KINGSTON 256 Mb
mother board: intel desktop board D865PERL
Pentium IV 2593 Mhz
2 Intel 100 Pro (2 Compaq PCI NC3121 Intel SB82558B)
3 3com 3C905CX-TX-NM
Average CPU Load 8
Max CPU Load 20
Crash seen before this : 0
Crash in this pop: 4
Bios Settings default , ide, halt on no errors, apc always on

Router B (The original router, currently offline)
IDE Drive 10 Gbs
RAM KINGSTON 256 Mb
mother board: intel desktop board D865PERL
Celeron 2.4 Ghz
2 Intel 100 Pro (2 Compaq PCI NC3121 Intel SB82558B)
1 3com 3C905C-TX ETHERLINK
Average CPU Load 10
Max CPU Load 20
Time working with out a crash 9 months
Crash in the last weeks : 6
Bios Settings default , ide, halt on no errors, apc always on

Cameron Earnshaw · Thu May 12, 2005 7:37 am

Are you doing any pcq or p2p queues (using mangle)?

Cameron

jmedinas · Thu May 12, 2005 3:16 pm

Are you doing any pcq or p2p queues (using mangle)?

Cameron

I don't use any p2p queues in this router and I have 70 Simples QUEUES.

regards...

randyloveless · Fri May 13, 2005 5:10 am

are you using sdram or ddr. if ddr what version.

one other thing are you using bridging ?\

i was told from MT support that if bridging , not run arp=disable on interface and run arp or proxy-arp from brige. this did seem to fix most of my hard crashes on a few of my systems .

they were all the same 3 like systems and all had same issue would just lock up watchdog would work couldnt get even the keyboard to work , had to hard reboot units before it would come back online.

randy

Router Crash Serveral times a Week

Router Crash Serveral times a Week

Who is online