So I have a cluster (2-node, no HA), and one of the nodes has error - and yet it works?

[Bamsefar]

So I have two nodes, pve and pve2. It is not identical hardware, and they are attached with a 2.5Gbit Ethernet. Most things look okay, but then there is two things I do not understand:
1) pve node seems to not find pve2 active link
2) pv2 node seems to get an error about quorum - but that is not in the pve node

I have attached journalctl and corosync status - any idea what to look for?

Edit: Oh and just to be perfectly clear: pve2 node is empty. I just installed Proxmox 9 on that box. So it has no guests at the moment, and well I could just delete that node - but that will not solve the challenge?

 

[Johannes S]

Why do you want to use a cluster? If you just want to be able to migrate VMs between them you could also use a container or vm with the datacenter manager for this and it would save you the hassle envolved with building a cluster and it's requirements (like dedicated network for corosync etc): https://pve.proxmox.com/wiki/Cluster_Manager#pvecm_cluster_requirements

 

[Bamsefar]

Well, I have no good answer to that, but the fact I still would like to learn. It is a homelab. And for what it is worth, the setup was worse since I had to add the pve and pve2 to etc/hosts to even get to this point, which is strange since I only gave it the IP addresses when I created everything. No hostnames. So it is a quest I guess?

 

[Bamsefar]

And Yes, I plan to add a 3rd node in my learning curve boll thing.

 

[Bamsefar]

To simplify (?) my questions, here are three direct questions:

1) In status of corosync I can read: loaded (/usr/lib/systemd/system/corosync.service; disabled - that is on pve node, on pve2 node it is enabled. Is there any reason for this?

2) In the same output just a few rows below:
[KNET ] host: host: 2 has no active links
[KNET ] link: Resetting MTU for link 1 because host 2 joined
[KNET ] host: host: 2 (passive) best link: 1 (pri: 1)

So there is no active link, yet it has a working link anyway?

3) In the journalctl output for the Corosync, and ONLY in pve2 node:
Sep 14 10:42:40 pve2 pmxcfs[1171]: [quorum] crit: quorum_initialize failed: CS_ERR_LIBRARY (failed to connect to corosync)
Sep 14 10:42:40 pve2 pmxcfs[1171]: [quorum] crit: can't initialize service
Sep 14 10:42:40 pve2 pmxcfs[1171]: [confdb] crit: cmap_initialize failed: CS_ERR_LIBRARY (failed to connect to corosync)
Sep 14 10:42:40 pve2 pmxcfs[1171]: [confdb] crit: can't initialize service
Sep 14 10:42:40 pve2 pmxcfs[1171]: [dcdb] crit: cpg_initialize failed: CS_ERR_LIBRARY (failed to connect to corosync)
Sep 14 10:42:40 pve2 pmxcfs[1171]: [dcdb] crit: can't initialize service
Sep 14 10:42:40 pve2 pmxcfs[1171]: [status] crit: cpg_initialize failed: CS_ERR_LIBRARY (failed to connect to corosync)
Sep 14 10:42:40 pve2 pmxcfs[1171]: [status] crit: can't initialize service

I have seen this in a few other posts, but no direct answeer to why this happens, and only on one of two nodes. Remember pve2 has no workload, no VMs/LXC or anything - and from a hardware point it is superior to pve (CPU/RAM/NVMe disc). And it has a dedicated 2.5Gbit ethernet. Talk about strange curveboll....?