Nutanix: Disable Dual Stack to register Prism Element into Prism Central
When deploying Prism Central on my Nutanix CE 2.1 cluster, I was faced with an error message that prevented registering my cluster on Prism Central…
Indeed, once all the parameters are filled in during validation, I get the error message “Cluster has dual stack enabled. Cannot register to a PC.” :
This error message is related to the presence of IPv6 in parallel with IPv4 on the cluster, but a solution exists and it resides in the following command:
manage_ipv6 unconfigure; manage_ipv6 disableYou must answer “Y” to the question “Proceed to remove above IPv6 configuration?” in order to validate the process:
nutanix@NTNX-436d2f97-A-CVM:192.168.84.200:~$ manage_ipv6 unconfigure; manage_ipv6 disable
[INFO] Initializing script… done
[INFO] Current IPv6 configuration on cluster: {
"svmips": {
"192.168.84.200": null
},
"hostips": {
"192.168.84.199": null
},
"prefixlen": null,
"gateway": null
}
[INFO] Note: This operation will restart the following services: ['CerebroService', 'StargateService']
Proceed to remove above IPv6 configuration? [Y/N]: Y
[+] CVM and Hypervisor IPv6 addresses unconfigured
[+] Cleared IPv6 configuration from Zeus
[+] CVM and hypervisor firewall rules updated
[+] Necessary services have been restarted
[INFO] Marked Ergon task 4673fda3-92da-4efe-59f5-1dd3fc51a6cd as kSucceeded
[INFO] Action unconfigure completed successfully
Script output logged to /home/nutanix/data/logs/manage_ipv6.out
[INFO] Initializing script… done
[+] IPv6 disabled on CVMs and Hypervisors
[INFO] Marked Ergon task 9cdfcb37-436a-479f-4d7a-08ef69e266b0 as kSucceeded
[INFO] Action disable completed successfully
Script output logged to /home/nutanix/data/logs/manage_ipv6.out
nutanix@NTNX-436d2f97-A-CVM:192.168.84.200:~$Once the procedure is completed, the Prism Central registration goes smoothly:
If you still have issue, Goerge bring us another way to disable dual stack :
Add all these lines in /etc/sysctl.conf:
net.ipv6.conf.all.disable_ipv6 = 1
net.ipv6.conf.all.disable_policy = 1
net.ipv6.conf.default.disable_ipv6 = 1
net.ipv6.conf.default.disable_policy = 1
net.ipv6.conf.eth0.disable_ipv6 = 1
net.ipv6.conf.eth0.disable_policy = 1
net.ipv6.conf.eth1.disable_ipv6 = 1
net.ipv6.conf.eth1.disable_policy = 1
net.ipv6.conf.eth2.disable_ipv6 = 1
net.ipv6.conf.eth2.disable_policy = 1
net.ipv6.conf.lo.disable_ipv6 = 1
net.ipv6.conf.lo.disable_policy = 1Run :
sudo sysctl -p And rolling reboot of all the CVMs.
Then run manage_ipv6 unconfigure :
manage_ipv6 unconfigure; manage_ipv6 disable
This solution is not working…
I get the following message when attempting this…
[INFO] Marked Ergon task as kFailed [-] Failed to run action unconfigure. Error: Failed to unconfigure IPv6 addresses
This solution works. I test everything beforing posting here, and it works like a charm for me.
Same here on 6.8.1 every single time every single deployment.
I believe you Julien and enjoy your content. I have been fighting this error for a long time. Multiple reloads and cannot seem to get past it. I have used the “script” method included in this tutorial and also disabled ipv6 manually. When I run the ifconfig it clearly shows no ipv6 addresses but the for some reason the Ergon task error is always the same. Not sure what else I can do. Currently running ESXI v 7.0 Update 3 for the host. Any additional suggestions you may have would be appreciated.
Oh your have deployed a nested Nutanix CE ? Is Prism Central already deployed ?
I have tried it multiple ways. I have installed CE and then done the install of the “small” Prism Central, but when I go to add the CE node to Prism Central the it pops up with the dual stack message. So then I go in to disable the IPV6 and get the Ergon error message when I use the manage_ipv6 disable tool. The only way I am able to disable IPv6 is through using the sudo sysctl -w net.ipv6.conf.all.disable_ipv6=1 command which is not persistant or adding the three lines to the /etc/sysctl.conf file. When I edit the sysctl.conf file it does appear that ipv6 is disabled after I reboot the cvm, but it still will not allow me to add it to PC because the dual stack error still shows up. For some reason the manage_ipv6 unconfigure; manage_ipv6 disable commands always fail with the same Ergon error message. I have also tried disabling IPV6 before I install Prism Central to see if that would help, but I still get the same error. I am thinking of reloading my ESXi 7 server (Dell Precsision 7920, 32 cores, 528GB ram, 4TB NVME) with Windows Server and then instaling VMware Workstation to see if that helps, but would like it better if I did not have to do that. Prism Central does install, but it does not automatically show the single node cluster. The error pops up when I attempt to add the node to the cluster. I have not attempted to add another single node cluster the PC instance on a different clust and will be attempting that also, but it would really be nice to have way to add it as simply as outlined in your posting. Many thanks for your response, sincerely appreciated.
Regards,
Eric
I was able to disable IPv6 on Nutanix CE CVM, just do sudo manage_ipv6 unconfigure; manage_ipv6 disable as the command and it works.
It appears that it does not work for version AOS 6.10 (at least that is in my case). Disable IPv6 across all CVMs, rebooted, made sure CVMs are clean of all IPv6, then try to connect PC, and it would work. In my case it did.
The version of AOS I see in my Prims Element is listed as ver 6.8.1 which is what came with the latest Nutanix CE install. Have you had any issues with this version? Also the version of PC I am installing is 2024.1.0.2 which is the only version that will install with Nutanix CE from my understanding and requires me to manually import it.
Fought through this today with CE 6.81. Like Eric, the manage_ipv6 disable command would consistently error out. I added the three lines for pre-AOS 6.5 from this page (https://portal.nutanix.com/page/documents/kbs/details?targetId=kA00e000000LJDbCAO) to each host and rebooted the cluster. While manage_ipv6 show indicated that there were no longer IPv6 addresses present for the internal nics, I still received the same dual stack error when attempting to register to Prism Central. Ran the manage_ipv6 unconfigure and disable commands again and it resolved the problem.
It eventually worked – what I did: first, I added all these lines in /etc/sysctl.conf , ran sudo sysctl -p and also did a rolling reboot of all the CVMs:
net.ipv6.conf.all.disable_ipv6 = 1
net.ipv6.conf.all.disable_policy = 1
net.ipv6.conf.default.disable_ipv6 = 1
net.ipv6.conf.default.disable_policy = 1
net.ipv6.conf.eth0.disable_ipv6 = 1
net.ipv6.conf.eth0.disable_policy = 1
net.ipv6.conf.eth1.disable_ipv6 = 1
net.ipv6.conf.eth1.disable_policy = 1
net.ipv6.conf.eth2.disable_ipv6 = 1
net.ipv6.conf.eth2.disable_policy = 1
net.ipv6.conf.lo.disable_ipv6 = 1
net.ipv6.conf.lo.disable_policy = 1
Then manage_ipv6 unconfigure worked and I could also register the cluster to PC. AOS 6.10.1 + AHV el8.nutanix.20230302.103003, pc.2024.3.0.2.
Thanks! This worked for me with Nutanix CE
This process will work fine:
1) SSH into the CVM nutanix@NTNX-3fcd56af-A-CVM:~$
2) nutanix@NTNX-A-CVM:~$: sudo nano /etc/sysctl.conf and add the following lines:
net.ipv6.conf.all.disable_ipv6 = 1
net.ipv6.conf.default.disable_ipv6 = 1
net.ipv6.conf.lo.disable_ipv6 = 1
3) nutanix@NTNX-A-CVM:~$: genesis restart >>> [RESTARTING GENESIS IS IMPORTANT]
4) nutanix@NTNX-A-CVM:~$: manage_ipv6 unconfigure; manage_ipv6 disable
5) Now the dual stack error should be gone, if not then restart the CVM
Optional: nutanix@NTNX-A-CVM:~$: sudo reboot -h now
nutanix@NTNX-05197cc2-A-CVM:192.168.1.121:~$ manage_ipv6 show [INFO] Initializing script… done
gateway: null
hostips:
192.168.1.120: 2001:569:54fe:1b00:8eec:4bff:feb2:b535
prefixlen: null
svmips:
192.168.1.121: null
[ERROR] Problems identified in the cluster:
[ERROR] 1. One or more CVMs missing an IPv6 address
[ERROR] 2. Mismatch found between IPv6 address configured on hypervisor 192.168.1.120 and Zeus
In my case, “hostips” is not “null”. IPV6 still exist. How do we put a “null”. Any help much appreciated.