Some thoughts for an HA homesetup on Proxmox
- for a "regular" HA cluster you'd need at least 3 physical machines. A cool Proxmox feature is the external quorum vote form a non cluster device, i.e. a daemon on some random server in your network. This allows an HA setup with only 2 machines (Qdevice https://pve.proxmox.com/pve-docs/chapter-pvecm.html#_corosync_external_vote_support ).
- the distributed Ceph filesystem can be used as cluster shared storage, eliminating SPoF of a "regular" shared storage (eg. NAS).
- easiest failover could be an automated migration of the node VM to another physical device. If zero downtime is mandatory a running standby node with fail over of data filesystems could be a scenario.
BLue
npub1ml…vkygu
2025-03-23 18:59:08
in reply to nevent1q…02ta
Author Public Key
npub1mlzz3vhzrv4j5rs2d70sun254qhw5s4l2s2amf6nx6c3epp844psevkyguSeen on
wss://nostr.winePublished at
2025-03-23 18:59:08Event JSON
{
"id": "05b04aa115b56513f6392b6ebe064417bebfbbff2595fadca348a27dc3a4b57e",
"pubkey": "dfc428b2e21b2b2a0e0a6f9f0e4d54a82eea42bf5415dda75336b11c8427ad43",
"created_at": 1742756348,
"kind": 1,
"tags": [
[
"e",
"9f91e340e825e5e9b4b6adf971bbda8c85ef981a6c38a2421c1e708bf7c790ae",
"",
"root"
],
[
"e",
"0e0b851142498e81dc55e32737f579a1f4c5a39fc78affcf746cbe0e7334eac9",
"",
"reply"
],
[
"p",
"74ffc51cc30150cf79b6cb316d3a15cf332ab29a38fec9eb484ab1551d6d1856"
],
[
"p",
"d164d78c0b58edeb22d54942ccf2f8155befa45dc0e3717cc1f81d0f113451bf"
],
[
"r",
"https://pve.proxmox.com/pve-docs/chapter-pvecm.html#_corosync_external_vote_support"
]
],
"content": "Some thoughts for an HA homesetup on Proxmox\n\n- for a \"regular\" HA cluster you'd need at least 3 physical machines. A cool Proxmox feature is the external quorum vote form a non cluster device, i.e. a daemon on some random server in your network. This allows an HA setup with only 2 machines (Qdevice https://pve.proxmox.com/pve-docs/chapter-pvecm.html#_corosync_external_vote_support ).\n\n- the distributed Ceph filesystem can be used as cluster shared storage, eliminating SPoF of a \"regular\" shared storage (eg. NAS).\n\n- easiest failover could be an automated migration of the node VM to another physical device. If zero downtime is mandatory a running standby node with fail over of data filesystems could be a scenario.",
"sig": "f6719588afe0430feca9d5940499c3a5d4ead78c4d4e98603fa596a4e4ff91d14e995db833be321b036fd62cac7f821ad08fc6a4c7e9e7d2cab2ea882017a3be"
}