Cisco 3850 Unit deleted from Stack

[joeashby16885]

I have had an issue with my stack and need to know the cause. The stack deleted one of the units from the stack and caused network to be disconnected does anyone know why this occurred? Switch 2 was disconnected. 

I have attached the logs 

[Mark Malone]

hi
what was the issue did you reload it ? some of the logs are pointing to the bug below , do you have support to TAC this you could get clarification but if teh stack has been rebooted it may have cleared the fault or process which caused the issue

https://bst.cloudapps.cisco.com/bugsearch/bug/CSCup49704/?rfs=iqvred


3850 FED Crash - Waiting for SPI channels FED_SPI_FLCD,FED_SPI_FAST ...
CSCup49704
Description
Symptom:
When abruptly reloading (eg: manual power cycle) or adding/removing a switch member from a 3850 stack, other switches in the stack may begin crashing on the FED (forwarding engine driver) due to a timeout on the SPI, which is the internal message-passing system that lets services communicate with one another.

The following logs may be seen for the stack members that enter a crash loop:
FED-3-INIT_FAILED MEMBER: X fed: Module SPI Channel failed initialization Waiting for SPI channels FED_SPI_FLCD,FED_SPI_FAST_CONV,
IOSXE-3-PLATFORM MEMBER: X process sysmgr: Service [fed] pid:[XXXX] terminated abnormally [6].

The bug is for 3850 and 3650 as its a timing issue and not related to any particular HW.

Conditions:
A stack member must be either be reloaded abruptly (eg: manual power cycle) or added/removed from the stack.

Workaround:
Reload the entire stack to resolve a sync issue.

Further Problem Description:
if your code already has the fix of this bug CSCup49704, then you might want to check out CSCvb75523

[joeashby16885]

Hi Mark 

Thanks for the speedy reply, it was called in as loss of part of the network I wasn't carrying out any works or on site. it just happened which is why I'm struggling to work out why? the stack wasn't touched and this occurred? 

[Mark Malone]

Hi check the show version see why it failed look for on each switch in show ver Last reload reason: PowerOn

also check each switch flash for any crash files , it could of hit a process and flipped the whole stack ,it happens rarely though , could of been external power factor too

best to check show version to start with see what the reload output states

[Leo Laohoo · Accepted solution]

---

@joeashby16885 wrote:

Version 03.06.05E

---

I know exactly what this is:  FN - 64191 - Catalyst 3850/3650 Switches That Run Cisco IOS XE Releases 3.6.5E/3.7.4E or Earlier Might Exhibit High CPU Usage - Workaround Provided

[joeashby16885]

Thank you this makes sense as the stack that had the issue has a run time of over 6 months. Do you know where I can get the latest firmware versions from to upgrade? 

thanks 

[Leo Laohoo]

Stick with the latest 16.6.X or 16.9.X.