After upgrading our 9800 WLC to 17.9.2, Prime has started indicating that the Coral connection is down. It was fine in 17.6.x. Prime is 3.10.2 with all of the latest patches. As far as I have been able to determine the actual telemetry stream is fine though. AP discovery completes with no issues, I have client details, etc.
This document and this bug both mention the command:
show telemetry internal protocol cntp-tcp manager x.x.x.x 20828
https://www.cisco.com/c/en/us/support/docs/wireless/catalyst-9800-series-wireless-controllers/214286-managing-catalyst-9800-wireless-controll.html
https://bst.cloudapps.cisco.com/bugsearch/bug/CSCvs40684
It appears that Prime uses that command in the "Catalyst 9800 Telemetry Coral Health" service check. However, as of 17.8.1+ that command requires additional arguments (It wants the vrf-id and mgmt IP). Prime doesn't use the VRF-aware command syntax though. I've tested with both 9800-80 and 9800-L-F models, I can't speak for any other 9800 model
wc9800-06#show telemetry internal protocol cntp-tcp manager x.x.x.x 20830 ?
source-vrf Source VRF
wc9800-06#
wc9800-06#show telemetry internal protocol cntp-tcp manager PrimeInfrastructure_IP 20830 source-vrf 0 WLC_MGMT_IP
Telemetry protocol manager stats:
Con str : PrimeInfrastructure_IP:20830:0:WLC_MGMT_IP
Sockfd : 109
Protocol : cntp-tcp
State : CNDP_STATE_CONNECTED
Table id : 0
Wait Mask :
Connection Retries : 0
Send Retries : 0
Pending events : 0
Source ip : WLC_MGMT_IP
Bytes Sent : 22288781
Msgs Sent : 20578
Msgs Received : 0
Creation time: : Sun Dec 25 16:32:27:746
Last connected time: : Sun Dec 25 16:32:27:747
Last disconnect time: :
Last error: :
Connection flaps: : 0
Last flap Reason: :
Keep Alive Timeouts: : 0
Last Transport Error : No Error
As a test to see if this was the only thing that is affecting that service, I made an EEM script that translates from the original command to the new syntax. With that script in place Prime's "Catalyst 9800 Telemetry Coral Health" service is happy. So it seems it was that single command's syntax changing that broke it. To be clear, this doesn't affect telemetry actually functioning, it only impacts Prime's reporting of the telemetry status.
event manager applet telemManagerCmd authorization bypass
event cli pattern "^show telemetry internal protocol cntp-tcp manager [[:digit:]\.]+ [[:digit:]]+$" enter
action 000 set wlc_mgmt_ip" "Unknown"
action 001 set wlc_mgmt_vrf_id "0"
action 002 set pi_ip "Unknown"
action 003 set pi_port "Unknown"
action 004 regexp "([[:digit:]\.]+) ([[:digit:]]+)" "$_cli_msg" ignore pi_ip pi_port
action 005 cli command "enable"
action 006 cli command "terminal length 0"
action 007 cli command "terminal width 0"
action 008 cli command "show wireless interface summary"
action 009 regexp "[[:alnum:]\/]+[[:space:]]+Management[[:space:]]+[[:digit:]]+[[:space:]]+([[:digit:]\.]+)" "$_cli_result" ignore wlc_mgmt_ip
action 010 cli command "show telemetry internal protocol cntp-tcp manager $pi_ip $pi_port source-vrf $wlc_mgmt_vrf_id $wlc_mgmt_ip"
action 011 puts "$_cli_result"
Is anyone else seeing this with 9800 WLC versions 17.8.1 and higher? I'm curious if there is something particular to our setup where it makes the older CLI command disappear or if Prime uses the VRF-aware command syntax somehow.
