Power outage = no more Telnet/SSH

Michael Hutt · ‎02-09-2009

Hello everyone,

my company had a power outage over the weekend that apparently lasted longer than the UPS' could handle, and several of our switches lost power. No big deal, I back up the running configs every night at midnight, but now I'm unable to SSH/Telnet in to two of my switches (they're physically on top of each other). They seem to be functioning fine otherwise, and I can access them via HTTP, they react to pings and the attached clients haven't called to complain about network access, but I can't figure out why SSH support is suddenly gone.

Anyone have any ideas? Would rebooting the switches help?

paolo bevilacqua · ‎02-09-2009

Someone has likely forgot to save configuration. You need to connect locally, reconfigure, save.

In critical applications it's also important to do a power failure test to see if everything come back up as expected.

Michael Hutt · ‎02-09-2009

That's just the thing... I remember saving the configs on Friday, which I can confirm by the fact that what I changed still works today, with the power failure in between. The only thing that's not how it should be is the SSH access...

paolo bevilacqua · ‎02-09-2009

Invariably once on site you will identify some detail.. It's not a known problem so you will need to do your investigations.

Leo Laohoo · ‎02-09-2009

Can you "cluster" into the two switches?

Presume the following:

Working switch = SwitchA

Notworking switch = SwitchB

Here's how it works (hoping you got the right IOS):

SwitchA#sh cdp neighbor

! To see if you can see SwitchB.

SwitchA#config t

SwitchA(config)#cluster enable SwitchA

! This line enables clustering on the working switch.

SwitchA#end

SwitchA#sh cluster candidate

! Look for SwitchB's MAC address.

SwitchA#config t

SwitchA(config)#cluster member mac password

! Takes about 3 seconds if all goes well. You'll get "password incorrect" if the password is wrong. If you get "Access Denied" or something ... can't be helped anymore.

SwitchA(config)#end

SwitchA#rc 1

! Remote Cluster to the "first" switch (aka SwitchB)

Hope this works.

Michael Hutt · ‎02-09-2009

I figured out what the problem was. Apparently the power outage and subsequent reboot caused the switches to chanve the access-class for the VTY lines, meaning they were set for an access-list that didn't exist.

I swear I don't know how that happened. I was on both switches two day before via SSH because I needed to configure ports for new clients there, and I saved the configs. Strange.