Ok, how do I create cluster?

[xsandman]

Ok, Hi.
Sorry for a newbie question.
I have 2 proxmox servers(version 4.1-1) in same network and NFS storage. How do I create 2-node cluster?

 

[dietmar]

see https://pve.proxmox.com/wiki/Proxmox_VE_4.x_Cluster

 

[xsandman]

Yeah, I have missed that. Somehow. Thanks.

 

[xsandman]

I did
root@pve2:~# pvecm create gdnetcl

and

root@pve2:~# pvecm status

brings me

Cannot initialize CMAP service

Any suggestions, please?

 

[xsandman]

also launch from terminal brings me this.

root@pve2:~# systemctl status corosync.service -l
● corosync.service - Corosync Cluster Engine
Loaded: loaded (/lib/systemd/system/corosync.service; enabled)
Active: failed (Result: exit-code) since Fri 2016-04-08 16:09:09 MSK; 21s ago
Process: 1430 ExecStart=/usr/share/corosync/corosync start (code=exited, status=1/FAILURE)

Apr 08 16:08:08 pve2 corosync[1441]: [SERV ] Service engine loaded: corosync cluster closed process group service v1.01 [2]
Apr 08 16:08:08 pve2 corosync[1441]: [QB ] server name: cpg
Apr 08 16:08:08 pve2 corosync[1441]: [SERV ] Service engine loaded: corosync profile loading service [4]
Apr 08 16:08:08 pve2 corosync[1441]: [QUORUM] Using quorum provider corosync_votequorum
Apr 08 16:08:08 pve2 corosync[1441]: [QUORUM] Quorum provider: corosync_votequorum failed to initialize.
Apr 08 16:08:08 pve2 corosync[1441]: [SERV ] Service engine 'corosync_quorum' failed to load for reason 'configuration error: nodelist or quorum.expected_votes must be configured!'
Apr 08 16:09:09 pve2 corosync[1430]: Starting Corosync Cluster Engine (corosync): [FAILED]
Apr 08 16:09:09 pve2 systemd[1]: corosync.service: control process exited, code=exited status=1
Apr 08 16:09:09 pve2 systemd[1]: Failed to start Corosync Cluster Engine.
Apr 08 16:09:09 pve2 systemd[1]: Unit corosync.service entered failed state.

 

[xsandman]

Well, I made it working. Thanks.

 

[Arthur Stephens]

Please share how... I am doing a band new install.

root@promox2:~# pvecm status
Cannot initialize CMAP service
root@promox2:~# systemctl status corosync.service -l
â corosync.service - Corosync Cluster Engine
Loaded: loaded (/lib/systemd/system/corosync.service; enabled)
Active: failed (Result: exit-code) since Wed 2016-04-20 12:08:22 PDT; 17min ago
Process: 2097 ExecStart=/usr/share/corosync/corosync start (code=exited, status=1/FAILURE)

Apr 20 12:07:21 PteraCluster corosync[2108]: [SERV ] Service engine loaded: corosync configuration service [1]
Apr 20 12:07:21 PteraCluster corosync[2108]: [QB ] server name: cfg
Apr 20 12:07:21 PteraCluster corosync[2108]: [SERV ] Service engine loaded: corosync cluster closed process group service v1.01 [2]
Apr 20 12:07:21 PteraCluster corosync[2108]: [QB ] server name: cpg
Apr 20 12:07:21 PteraCluster corosync[2108]: [SERV ] Service engine loaded: corosync profile loading service [4]
Apr 20 12:07:21 PteraCluster corosync[2108]: [QUORUM] Using quorum provider corosync_votequorum
Apr 20 12:08:22 PteraCluster corosync[2097]: Starting Corosync Cluster Engine (corosync): [FAILED]
Apr 20 12:08:22 PteraCluster systemd[1]: corosync.service: control process exited, code=exited status=1
Apr 20 12:08:22 PteraCluster systemd[1]: Failed to start Corosync Cluster Engine.
Apr 20 12:08:22 PteraCluster systemd[1]: Unit corosync.service entered failed state.

 

[xsandman]

1. At the beginning of the installation process you should set up ip addresses, that you're NOT going to change later. In my case corosync(/etc/corosync/corosync.conf) was getting 172.16.1.77 from somewhere even after i edited /etc/network/interfaces and that caused troubles with corosync service.
2. Make sure all nodes know each other by names (proxmox1, proxmox2, etc.. in your case)
3. Also /etc/corosync/corosync.conf should have master ip adrress at bindnetaddr at "totem" part on ALL nodes.
I'm not sure I remember all the details, but these are main.
Hope you'll deal with it.

 

[fibo_fr]

There is this delicate time where you have just one server working... and the cluset somehow expects a quorum of 2
use "pvecm e 1" to tell it that the expected quorum is 1, not 2. You should then have wtite access to /etc/pve.

If this dir is empty or read-only, your secret weapon is to use "proxmox file system" (pmxcfs) in mode "local" rather than "cluster wide": issue the command "pmxcfs -l" (lower case L as local)

Just in case: you might have a look at the (non-validated) page I wrote on the wiki about corosync.conf and pmxcfs:
http://pve.proxmox.com/wiki/Editing_corosync.conf

 

[Arthur Stephens]

Yup. Must have done some thing when I first installed. I installed again and immediately went ssh to create the cluster. It is working now.
Thanks