topic Re: с3845 problem with multicast in Switching

с3845 problem with multicast

nivic — Wed, 06 Mar 2019 22:42:55 GMT

On cisco 3845 ( 12.4(25d) with EtherSwitch module I see constantly next messages:

"Insert multicast address to MARL table failed"

Then Router begin don't answer on console access and then router crash and reboot

%ALIGN-1-FATAL: Illegal access to a low address 11:51:04 MSK Mon Dec 27 2010
 addr=0xA8, pc=0x6072CA30 , ra=0x6072CA14 , sp=0x62EF9E80

%ALIGN-1-FATAL: Illegal access to a low address 11:51:04 MSK Mon Dec 27 2010
 addr=0xA8, pc=0x6072CA30 , ra=0x6072CA14 , sp=0x62EF9E80


 11:51:04 MSK Mon Dec 27 2010: TLB (store) exception, CPU signal 10, PC = 0x6072CA30


Is it software BUG or no ?

Re: с3845 problem with multicast

Peter Paluch — Mon, 27 Dec 2010 10:27:11 GMT

Hello,

Most probably, this is a software bug. Do you have an option of trying a different IOS version? The 12.4(25d) is fairly recent as far as I know - therefore you may want to try an older build.

If you have a support contract then I strongly suggest you raise a TAC case because this is certainly something that needs to be taken care of by the development team. We may find a workaround but a definitive solution can be made only by people developing the IOS. A TAC case can be raised directly from this thread.

Best regards,

Peter

Re: с3845 problem with multicast

nivic — Mon, 27 Dec 2010 10:44:21 GMT

Thank, for your quik answer.

I also think that this bug. I tried the early releases of IOS, the router crashes with same error.

Re: с3845 problem with multicast

Peter Paluch — Mon, 27 Dec 2010 10:55:24 GMT

Hello,

Hmmm... It seems that this bug has actually been discussed on these forums before:

https://supportforums.cisco.com/thread/344750

Unfortunately, no solution was found neither. Can you post the complete configuration of your device (the passwords and IP addresses sanitized, of course)? It seems to be somehow related to the presence of the EtherSwitch module - that module was used in both cases. In any case - if this is indeed a bug which it seems to be then I do not expect to be successful even after seeing your config, but then again, it would be a waste of chance to not to try verifying it.

Best regards,

Peter

Re: с3845 problem with multicast

nivic — Mon, 27 Dec 2010 11:27:21 GMT

my config below

I fine next bug - CSCsz89093

http://tools.cisco.com/Support/BugToolKit/search/getBugDetails.do?method=fetchBugDetails&bugId=CSCsz89093&from=summary

Related Bug Information

Multicast traffic is process-switched when egress int is GE-DCARD-ESW

May be I have similar problem ?

Re: с3845 problem with multicast

Peter Paluch — Mon, 27 Dec 2010 12:05:43 GMT

Hello,

I am not quite sure if your problem is related to the bug you have found. Your configuration does not contain any multicast-related statements - in other words, your router is not performing any multicast routing. The switching module itself may be performing IGMP Snooping (I suspect it is the default setting) but that is probably all. Are you aware of using any multicast application in your network?

Best regards,

Peter

Re: с3845 problem with multicast

nivic — Mon, 27 Dec 2010 12:31:45 GMT

Hello,

You are correct - we not use any multicast routing and multicast App in our network for business

Re: с3845 problem with multicast

Michael Simon — Sat, 01 Jan 2011 20:25:32 GMT

CSCsz89093 does not explain the symptoms. It only explains multicast traffic being dropped.

This issue is what we call a Bus Error.

The Alignment Errors indicate the system was trying to access an address of:

addr=0xA8

This is in the lower 16K of memory and no Cisco router has memory in this region.

TLB (store) exception, CPU signal 10, PC = 0x6072CA30

This is the actual crash cause.  It was a bus error (signal 10).  

This is also indicated by the "TLB (store) exception' implying it was trying to 
write to an inappropriate address. 

The most common cause of this crash is a software bug. 

By definition a bus error is one of three kinds of failures: 

1. accessing a memory address at which there is no memory

2. writing to a 'read only' memory address

3. Correctly reading from or writing to a valid memory that has a failure. 

The first two are software failures. The first is by far the most common.  

The third is a very rare occurrence.  

In the show context of the crashinfo there will be a value for BadVaddr. 

This is the address the system tried to acces that caused the crash.  

If this address is valid for a write in 'show region' the crash may be a 
hardware issue.

Otherwise it is a software failure and should be investigated by the TAC.  

In so new an IOS image we would want to see if there is an existing bug 
or it is a new issue. 

........Mike