Cisco 3845 router doesn't start. Reloading....

[Timocha]

What What could be the issue?

Physically the router is up, but  IOS is reloading...

please see an attached file.

[Joseph W. Doherty]

Looks to me like router is in a loop attempt to upgrade the ROMMON which fails, which causes the next attempt.

What's the history of this router?  I.e. anyone trying to upgrade ROMMON on it?

Unsure, without some research, whether you can override this such as by interrupting boot process and resetting config register or some such.  (If router was under maintenance, which I doubt it is, this is a good kind of issue to contact TAC.)

[Timocha]

yes you are right it is in a loop..  i bought it for my lab from eBay could you help me with interrupting boot process and resetting config register? i cant find  info on google.

thanks in advance!

[Joseph W. Doherty]

See if this helps:

https://community.cisco.com/t5/switching/cisco3845-cant-break-into-ios/td-p/1906626

https://www.cisco.com/c/en/us/support/docs/routers/3600-series-multiservice-platforms/15080-recovery-c3600.html

PS:

BTW, often a problem with used Cisco equipment on eBay, it's not in good working order.

[Timocha]

nothing was helped until i shut down the router, unplug two switchport blades and two  layer 3 fiber blades. Reload the router and vualia..

now the router booting without any issues. I plug back all extra blades i have removed before and router still working fine

So what was the problem i don't know, but i have 100% working router now thanks!!!

[Joseph W. Doherty]

Wonderful news!

From what you described, the "fix" might have just been due to reseating those boards.  What ROMMON now shows with a show version?

[Richard Burts]

I agree that there was a loop. But it was not an attempt to upgrade the ROMMON that caused the loop. rommon processing completed and it moved to to attempting to load IOS. The important thing in the output was the message about watch dog timeout reset. I agree that it is likely that reseating the cards was the solution. Glad that it is working now.

[Joseph W. Doherty]

@Richard Burts it's certainly possible I don't correctly understand what OP attachment shows, but I thought it unusual that each loop occurrence has:

Upgrade ROMMON initialized

Don't misunderstand, I agree the watch dog error appears to happen during the main IOS boot, but was there a problem with a ROMMON upgrade that caused the timeout for the IOS boot?

System Bootstrap, Version 12.4(13r)T5, RELEASE SOFTWARE (fc1)
Technical Support: http://www.cisco.com/techsupport
Copyright (c) 2007 by cisco Systems, Inc.

*** Watch Dog Timeout Reset ***

ra = 0x603876c4, EPC = 0x60fee900

*** BCM 1125 went through a full reset. HT/PCI fabric reset. ***

Total memory size = 512 MB - DIMM0 = 256 MB, DIMM1 = 256 MB
c3845 platform with 524288 Kbytes of main memory
Main memory is configured to 72/72(dimm 0/1) bit mode with ECC enabled


Upgrade ROMMON initialized

With resetting cards fixing the issue, and with the repetition of the ROMMON upgrade message, I wondered if ROMMON was unable to successful update something on one of the cards, which during the follow-on IOS boot, caused the watch dog timeout.  Ideally, I would love to have seen the first successful boot log, but also wondering whether current boot logs show the ROMMON upgrade message still.

If the ROMMON message no longer appear in the boot log, would you still be certain that a ROMMON upgrade was not the cause of the loop?

BTW, I'm not thinking, if you might be, of an all or nothing ROMMON upgrade because I once encountered a case where we had a 3750 with an issue after an IOS upgrade, which was done remotely (w/o console remote access) and we thought it all good as the 3750 booted into the new IOS.  However, finally got a remote console connection on it, and what was happening, on every boot, IOS was trying to still upgrade some ROMMON but failing because it would run out of space, in flash, as it expanded that particular ROMMOM patch code.  Fix was obtained by freeing up some additional flash space, rebooted again, ROMMON patch worked, next reboot, no new ROMMON updates shown being applied.

[Richard Burts]

@Joseph W. Doherty that is an interesting case with the rommon upgrade. If I am understanding correctly in your case there were console messages about the failure (it is just that time elapsed before they were noticed). In this case there are console messages and no mention of rommon failure. 

We will not know for sure what was the cause of this issue. It does seem that the root of the problem was something about those cards which reseating solved. And perhaps it is not so important whether it was a rommon thing or an IOS thing.

[Joseph W. Doherty]

"If I am understanding correctly in your case there were console messages about the failure (it is just that time elapsed before they were noticed)."

Correct.  Also we routinely upgraded Cisco devices without any console output, remote or local, also including, in this case, many other 3750s.  Literally, the IOS upgrading performing the ROMMON patching only needed a small amount of temporary flash, but on one particular 3750, its flash was almost completely filled.  (I don't recall release notes mentioning need for temporary flash space during the first boot of the upgraded IOS.  Usually if you have enough space to load the software, in flash, you're good to go.)

"In this case there are console messages and no mention of rommon failure."

Correct too.  One would expect some kind of error message, but depending on the nature of the error, only Cisco could clarify whether an error message should have been logged.  (If so, possibly also a "bug", but situation might be so unusual, such a bug might never be noticed.)

"We will not know for sure what was the cause of this issue."

Yep, we don't know cause, only can "see" there was a boot loop after an IOS loading watch dog timer timeout.

"It does seem that the root of the problem was something about those cards which reseating solved."

That seems most likely.  It's also possible, although more unlikely, if a boot was tried with cards removed (?), and if some ROMMON upgrade was being attempted, it might have "worked" without those cards in device.

"And perhaps it is not so important whether it was a rommon thing or an IOS thing."

Agreed.  Possibly biggest takeaway was removing and reseating cards.  An approach which can be tried, but I didn't even think of suggesting it.