As you could read none of my earlier posts I reconfigured my Nutanix CE node in a different vLAN and everything worked like a charm.
Worked? After all the configuration was done I moved my node back in its place in the technical cabinet and started the node so I could fiddle with it again. Logging into Prism Central gave me no issues, but when I tried to connect to Prism Elements the following message popped up:
What’s this? As long as Pulse is not working I cannot connect to the GUI, so this is annoying. I connect to the CVM using SSH to start troubleshooting.
- Check if DNS is working correct
nslookup insights.nutanix.com
2. Check if the required ports are not being blocked.
nc -vz ncs01.nutanix.net 80
nc -vz ncs01.nutanix.net 8443
nc -vz ncs02.nutanix.net 80
nc -vz ncs02.nutanix.net 8443
nc -vz insights.nutanix.com 443
3. Check if my CVM gets a response from insights.nutanix.com.
curl -k https://insights.nutanix.com:443
4. Run Nutanix Cluster Checks for pulse
ncc health_checks pulse_checks run_all
Last thing to check if the NTP configuration is still ok
ncli cluster get-ntp-servers
ntpq -p
hmm, all things look good, no issues.
Luckily I’m not the only one with this issue and I found the following on Reddit, stating this is an issue with AOS 6.8 including a workaround.
So in the CVM I make a backup of the cron_servicaebility.py first
sudo cp /home/nutanix/serviceability/bin/cron_serviceability.py /home/nutanix/serviceability/bin/cron_serviceability.bak
Now I can safely edit the file using vi.
vi /home/nutanix/serviceability/bin/cron_serviceability.py
Scroll down and search for the following text (line 607 for me):
if((svccfg.enable_default_nutanix_email()or
svccfg.enable_default_email_alert_contact_list()) and mode in ["leader"] and (not utils.smtp_server())):
and replace this with the following text
if(mode in {“leader”] and (not utils.smtp_server())):
And save the file.
run the script to check for syntax errors
/home/nutanix/serviceability/bin/cron_serviceability.py
Wait for a couple of minutes and login back to Prism Element. Now I see all the green check-marks and can login back to the cluster again.
Thanks @Gurft / Kurt Telep forthe solution, it drove me nuts 🙂