summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authoruvok cheetah2023-08-04 17:18:58 +0200
committeruvok cheetah2023-08-04 17:18:58 +0200
commit6fa729a5583b703ae5e459b528a1965ce839c3fd (patch)
treef8587af389ba230b5df427661f6b60e0150e904d
parent0312a291af136c5ccbbe7cff4852690f3d9d841c (diff)
Add proxmox article
-rw-r--r--_posts/2023-08-04-upgrading-my-proxmox.md39
1 files changed, 39 insertions, 0 deletions
diff --git a/_posts/2023-08-04-upgrading-my-proxmox.md b/_posts/2023-08-04-upgrading-my-proxmox.md
new file mode 100644
index 0000000..b1c6109
--- /dev/null
+++ b/_posts/2023-08-04-upgrading-my-proxmox.md
@@ -0,0 +1,39 @@
+---
+layout: post
+title: Upgrading my Proxmox
+date: 2023-08-04 17:03 +0200
+lang: en
+categories: tech
+---
+
+So yeah, today I was [upgrading my Proxmox from v7 to v8](https://pve.proxmox.com/wiki/Upgrade_from_7_to_8).
+
+Due to the recommendations in their wiki I decided against doing a live upgrade
+and instead went with reinstalling the new version.
+
+So... What do I do with the existing machines in the meantime? I could just do a backup and
+restore... But I wanted to try out the cluster thing. Reasoning was "if something goes wrong,
+I still need my DNS server" (which runs in a Proxmox container). With the cluster
+feature, I can move the VMs / containers to a different PC.
+
+So, I first set up a VM on my Linux PC (with libvirt / KVM / QEMU), installed Proxmox
+there, created a cluster from my current "main Proxmox", and joined the new machine.
+All via the web interface, which went very neat.
+
+After I migrated the containers over, I shut down my "main Proxmox" and wanted to
+remove it from the cluster (to avoid problems when re-adding it later).
+So I followed [the Wiki](https://pve.proxmox.com/wiki/Cluster_Manager),
+ssh'ed to my now "backup Proxmox", happily wanted to execute `pvecm delnode srv-proxmox`
+with the result of ... `cluster not ready - no quorum?`. Huh? Wait, what?
+
+Well, turns out, Proxmox by default seems to assume enterprise setups, and no enterprise
+has just two machines ;-). No, I didn't read about the specifics, but apparently you need
+at least two online machines at any given time when you have a cluster. I looked it up,
+after executing `pvecm expected 1` I was able to remove the soon-to-be-upgraded machine from
+the cluster.
+
+So, the installation itself went smooth. After the installation, I joined my "main Proxmox"
+back into the cluster, ssh'ed into it, executed a `pvecm updatecerts` as mentioned in the Wiki.
+
+Currently, I still have a cluster, although I may as well dissolve it, have to look up how.
+The second machine in a VM will go. I have no need for it anymore... until the next upgrade.