Not able to create 3 node cluster

  • 21 August 2019
  • 12 replies

#cluster -s $CVM_IP create
2019-08-21 18:53:02 INFO cluster:2638 Executing action create on SVMs $CVM_IP
2019-08-21 18:53:06 CRITICAL cluster:922 Could not discover all nodes specified. Please make sure that the SVMs from which you wish to create the cluster are not already part of another cluster. Undiscovered ips : $CVM_IP

I am getting error above, and this is fresh new install. None of the server has cluster running.

12 replies


I have the same issue that is because that there is no mdns in your environment.

If you make sure that your CVM can connect to each other well, try skipping discovery.


cluster -s IPaddress --skip_discovery create


Same issue? Any one give the root cause and how to fix  it ?

I am trying to install and configure three node cluster on dell hardware.
Userlevel 1
Badge +6
Struggling to know why you would not use the foundation process if this is a cluster new hardware purchased for Nutanix ?

If this is the case you should open a case with Nutanix support.

Or is not the case and you are trying to install the free version Nutanix CE on some hardware you have laying around ?

@ronnie I don't think so, I downloaded the image and prepared usb, attached the usb on all the three servers, boot the server using the usb and triggered install command to initiate the installation process as per the documentation.
Userlevel 1
Badge +6
Did you complete the install using the foundation process ?

It takes a couple of hours to rebuild everything and it creates the cluster after the build is complete.

If for some reason the foundation will not pick up the individual nodes you can add the mac addresses and it will force a rebuild and wipe and build each node.
@ronnie Sorry I did not get the part "much easier to just run foundation again as a bare metal install." do you mean re install everything? I just did reinstall recently.

this is a three node cluster that I was trying to setup. I am able to ping and ssh from all three cvm but when I try to create cluster it always says undiscovererd IP
Userlevel 1
Badge +6
If this is a new cluster it would be much easier to just run foundation again as a bare metal install.

I see CE mentioned is this just something you found as a fis or is it a three node CE cluster ?
@ronnie this is a new cluster, and I can ping and ssh to all the cvm in the cluster. I restarted Genesis on all the CVM too.

last time I manually triggered the zookeeper_init script to start zookeeper and added entry to the /etc/hosts, but after that foundation stops. So I destroyed everything and restarted, but again it is failing on same part.

I feel like this is the reason it is not working.

result: True, message = None
2019-08-27 14:51:20 INFO genesis is attempting to connect to Zookeeper
2019-08-27 14:51:40 ERROR Could not reach zookeeper
2019-08-27 14:51:40 ERROR Failed to get CVM id
2019-08-27 14:51:40 INFO Waiting for node to become configured through RPC

I found out some solution that I want to try as mentioned in

If this happens, use the following procedure to recover without needing to reinstall CE.

  1. Run the clean_disks command.
nutanix@cvm$ run clean_disks -p /dev/hard_disk_id1
  1. where hard_disk_id is the name of the disk in CVM. (Only disks that are not CVM boot drives need to be cleaned).
  2. Restart Genesis.
  3. [code]nutanix@cvm$ genesis restart[/code]
But how to find out which hard_disk is it to run the clean_disk, (ssd?)

Thank you for your help. 🙂
Userlevel 1
Badge +6
Is this a new build ?

Have you checked cluster status to make sure it's not part of a cluster already ?

can you ping each CVM ?
@ronnie that is just a sample of me saying CVMIP, since the ip is in corp network I replaced ip with $CVMIP
Userlevel 1
Badge +6
$CVM_IP needs to be replaces with the three IP addresses or the CVM's in the cluster

cluster -s,, create