noVNC ready: native WebSockets, canvas rendering” on VM screen while launching from Prism central

Problem Description :-

Virtual Machine console with Prism Central.

“noVNC ready: native WebSockets, canvas rendering” on VM screen while launching from PC”

Plan of Action

We need to schedule the upgrade to PC 5.17.1.1 which has the fix for the issue tracked in ENG-288461

[wpinsertinpostad id=”ylcxs5f10a8328948b”]
image 2
Virtual Machine Console Nutanix

Troubleshooting History :-

– The fix for the issue wasn’t available earlier, so a script was used for a temporary workaround so that whenever acropolis connections max out, it’ll restart the service on one CVM to resolve the issue automatically.
– This was performed around Mar 2020, which got rolled over due to the AOS upgrades in June 2020, as cronjob are non-persistent over upgrades
– Confirmed that we were hitting the same issue again and recommended upgrading PC to 5.17.1.1 which has the fix
– Thus restarted acropolis in Node .21 which was having 128 connections
– Checked and confirmed that we were able to launch the VM console after that.

Snippets


nutanix@NTNX-XXX-A-CVM:X.X.X.21:~$ ncli multicluster get-cluster-state


Cluster Id : ac26d794-e2c0-44b4-aa53-39d2bc364fd7

Cluster Name: Unnamed

Is Multicluster: true

Controller VM IP Addre... : [X.X.X.73]

External or Masqueradi... :

Controller VM NAT IP A... :

Marked for Removal: false

Remote Connection Exists: true


nutanix@NTNX-XXX-A-CVM:X.X.X.21:~$ nc -v X.X.X.73 9440

Ncat: Version 7.50 ( https://urldefense.proofpoint.com/v2/url?u=https-3A__nmap.org_ncat&d=DwIFaQ&c=s883GpUCOChKOHiocYtGcg&r=Vb_G-NyjEWHi-ESC7cC3g_8HKo9dQWOoPeqtvCBgpFQ&m=iWRUT2MmYvqN0ire7XtwGksypMBvGIyG0QtvPy8WXVg&s=tA8pZwOPUi8ojFfvM1WZ_yGfeeIdlARKqFDk7hJ26q0&e=  )

Ncat: Connected to X.X.X.73:9440.
nutanix@NTNX-XXX-A-CVM:X.X.X.21:~$ nuclei
2020/07/20 12:06:45 Connected to X.X.X.23:9876
2020/07/20 12:06:45 Authenticating connection 0x0
2020/07/20 12:06:45 nuclei are attempting to connect to Zookeeper
2020/07/20 12:06:45 Authenticated: id=0x2727bdddc9d6517, timeout=20000
remote_connection.health_check_all
"Number of remote connections in cluster are - 1"
"---------------------------------------"
"Performing health check on remote connection pe_00058a92-584d-4c91-0000-000000028462_pc_connection"
"Health check returned OK for given RC which has uuid:f6bb255f-1751-5880-8bd5-5f3f10745569 for {\"major_version\":\"3\",\"minor_version\":\"1\"}\n"
"---------------------------------------"
"Health check complete."
nutanix@NTNX-XXX-A-CVM:X.X.X.21:~$ for i in svmips; do echo $i; PID=$(ssh nutanix@$i ps -ef | grep "B /home/nutanix/bin/acropolis" | grep -v grep | awk '{print$2}'); ssh nutanix@$i "/usr/sbin/lsof -p $PID -nP | grep 2030 | grep ESTABLISHED | wc -l"; done
X.X.X.21
FIPS mode initialized
Nutanix Controller VM
FIPS mode initialized
Nutanix Controller VM
128
X.X.X.23
FIPS mode initialized
Nutanix Controller VM
FIPS mode initialized
Nutanix Controller VM
7
X.X.X.25
FIPS mode initialized
Nutanix Controller VM
FIPS mode initialized
Nutanix Controller VM
3
Upgrade history :
Thu, 04 Jun 2020 00:22:10 el7.3-release-euphrates-5.15.1-stable-4fbdd4d9de331230bb468b3549f530e80ab53bb9 - LTS
nutanix@NTNX-XXX-A-CVM:X.X.X.21:~$ allssh crontab -l | grep servicerestart.sh
nutanix@NTNX-XXX-A-CVM:X.X.X.21:~$
After Acropolis restart on .21:
nutanix@NTNX-XXX-A-CVM:X.X.X.21:~$ for i in svmips; do echo $i; PID=$(ssh nutanix@$i ps -ef | grep "B /home/nutanix/bin/acropolis" | grep -v grep | awk '{print$2}'); ssh nutanix@$i "/usr/sbin/lsof -p $PID -nP | grep 2030 | grep ESTABLISHED | wc -l"; done


X.X.X.21

FIPS mode initialized

Nutanix Controller VM

FIPS mode initialized

Nutanix Controller VM

0

X.X.X.23

FIPS mode initialized

Nutanix Controller VM

FIPS mode initialized

Nutanix Controller VM

7

X.X.X.25

FIPS mode initialized

Nutanix Controller VM

FIPS mode initialized

Nutanix Controller VM

3

RELEASE NOTES Prism Central 5.17.1.1

  • PC might fail to launch the PE web console of clusters registered with that PC.
  • Alert policies are not disabled by using the PE and PC web console
  • If you are a user with the role User IN PC, you cannot transfer the ownership of the VMs you have created to other users.
  • None of the entities specified in the recovery plan are recoverable.
  • cluster outage due to Zookeeper becoming unresponsive
  • PC might run out of disk space as multiple NGT Tool installers are stored in the home directory /home/ngt of PC.
  • If PC is destroyed or otherwise deleted, you might not be able to unregister a previously managed cluster
  • Improved self-help animations within the LCM UI.
  • Alert policies are not disabled by using the PE  and PC web console
  • UEFI boot is not supported for PC  VMs Workgoround is to create VM with the legacy mode in PC or PE UI or ACLI.
  • It Supports LCM 2.3.2 (Life Cycle Manager).
  • Upgrading PC from within Prism Central LCM 2.3.2
  • In-VM update improves disk and HBA firmware upgrade performance
  • For reducing host reboots, Batch upgrades combine applicable upgrades into fewer upgrade processes.
  • 1-click infrastructure upgrades with comprehensive dependency management
  • Automated All tasks with Prism Central and Prism Element Latest LCM version.

Leave a Reply