Unable to access 3850 switch stack via serial port or ssh

[steve switzer]

Hi

We have had a 3850 stack which was inaccessible via ssh

so logged in via a serial port to clear the vty lines

This was done with the 'clear vty 1 etc  command

but still unable to get in on ssh

When we tried it again we now found that we were

unable to access the serial connection

All that happens is we scroll the page

Anyone else had this

Steve

[Madhukrishnan Gopinathan Nair]

Not seen something like this before specillay scroll the page.

If it is comepletely in accessible, may be you can consider reloading the box

Thanks,

Madhu.

[ogmis1r]

I have this problem too. c3850 stack running 03.03.03. Can't login via SSH. I connected to the console and logged in. 'show users' says all vty connections are used. I tried the command  'clear line xx' and 'clear line vty xx'. these did not work.  A 'show processes cpu sorted' did not reveal anything and the stack seems to be running normally. Then I did a 'show run'......... and now I do not have console access.

This is not the first time i've come across this. Whilst a stack reboot will fix it, it seems a drastic measure.

Rob.

[daniel.yuste.aroca]

I have the same problem as ogmi1r.

Suddenly I was unable to login via SSH (SSH access to the switch has been working fine for months) , so I connected a console cable, executed "show log" and got the output whithout problems. Then I executed  "show run" and the console access become unresponsive too.

The stack seems running normally: LEDs are normal, the switch management IP is reachable via ping, users connected through this stack have normal connectivity.

My stack is running 03.06.01E

Have you been able to get a cause for this behaviour?

[ogmis1r]

Bug ID CSCuv69297

Fixed in  3.6.3

[gary olson]

Had the same issue with a switch stack running 3.6.3. I power-cycled the stack this morning and the stack is accessible now. I also upgraded to 3.6.4 which Cisco has starred as the recommended version.

[bjmayes01]

I've recently seen the problem resurface in the 3.6.3 code for 3850s too. It's not as often as the previous codes but it does happen. You can access the switches via console, just don't issue a "sh run", or try to save any configuration changes. Doing so will immediately lock the console session. If you console into a switch and issue a "sh user" on it, you'll notice that every vty line will be used and the switch ignores commands to clear them.

[ogmis1r]

Hi Madhukrishnan Gopinathan Nair,

We are a Hospital running c3850 stacks. Reloading the switches to fix management access while everything else appears to work, is not an option nor an acceptable solution.

Can you confirm this is fixed in 3.6.3 or provide the bug ID?

Thanks

Rob

[lateefwalana]

We have the same issue, running 03.03.05SE

[bjmayes01]

This is a code bug. We had this problem on 3.3.x code as well. We've been running the 3.6.3 code with no issue.

[jmartinez41]

we've seen it in 3.3.x, so we upgraded to 3.6.x, 3.7.x, now we're on 16.3.7 and we just had a dozen switches all do the same thing. Because we are looking at bringing DNA, I am about to upgrade all of the 3850's to 16.6.5 (Everest) because you need to be at least at 16.6.3 for stacked devices to be on DNA. I suspect, however, that the issue will not go away.