Proxmox cluster (3 nodes) unreachable GUI after power outage

[ctlister]

I just got back from a power outage, and after restarting all three nodes, none of my Proxmox nodes are reachable. I cannot access the web GUI nor SSH into 2 of the three nodes.

But connecting a VGA monitor I can see the console, and it does have the correct IP address.

What should I do?

 

[bbgeek17]

sounds like your cluster did not come up properly. I'd start with troubleshooting steps on this page: https://pve.proxmox.com/wiki/Cluster_Manager
search for "troubleshooting". Look at journalctl and various pvecm outputs.

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[ctlister]

I tried releasing the IP address on bond0 and then running dhclient -v bond0 and it says "no buffer space available". Before this, I was not able to ping my home router.

Edit: I just realized, that both nodes #1 and #3 utilize NIC-teaming through LACP. But node #2 can be SSHed into (but can't be reached with a GUI). It's possible that the Zyxel managed switch (which had really poor documentation on NIC-teaming setups) broke on me.

 

[bbgeek17]

are you using wifi adapter? If so - thats not something that is well supported. Perhaps a reboot of entire PVE environment might be helpful.
You should read through troubleshooting steps mentioned earlier.

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[ctlister]

I'll get it done after finals and before my doctor's appointment later this week. I have a good feeling it is my managed switch.

Edit: It is my managed switch. Brought Node #3 online that has NIC-teaming. Node #1 still isn't working. So during the weekend i'm going to restore to default network interfaces and put together NIC-teaming from scratch. The Zyxel switch is buggy, because the first time I implemented NIC-Teaming, I didn't need to use a LAG group, it just causes a DoS, but now putting together LAG groups for Node #3 works.

Fixed. My Zyxel Managed Switch dislikes my Dell R710 Rackserver with three bonded NICs and one sniffer NIC for SecurityOnion. Apparently I can't give it a LAG group for that. But it has taken a liking for my Dell Precision's bonded NICs.

I'm going to toss this Zyxel away for my Cisco Catalyst during winter break.