mirror of
https://github.com/proxmoxer/proxmoxer.git
synced 2026-04-25 07:06:00 +03:00
[GH-ISSUE #50] Question: do proxmox nodes share auth token? #23
Labels
No labels
backend:https
backend:local
backend:openssh
backend:ssh_paramiko
pull-request
status:ansible-issue
status:help-wanted
status:info-needed
status:proxmox-issue
status:review-needed
type:bug 🐞
type:dependency ⛓️
type:docs 📝
type:enhancement ⏫
type:maintenance 🛠️
type:meta
type:question ❓
type:request ✋
type:testing 🧪
version:1.x
version:latest
version:py2
version:py3
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
starred/proxmoxer#23
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @variable on GitHub (Apr 23, 2021).
Original GitHub issue: https://github.com/proxmoxer/proxmoxer/issues/50
Let's say I have setup a proxmox api load balancer that reverse proxy to my cluster.
If the load balancer switch to another node, does the existing auth token still work?
@jhollowe commented on GitHub (May 21, 2021):
yes, if your nodes are in a cluster, load balancing across their API will work.
I tested this by creating 2 nodes (in VMs) and joining them together in a cluster. I then created a container on one node, installed nginx, and added the following to
/etc/nginx/nginx.conf:using the container's IP address, I can access the web UI or use proxmoxer and the calls are evenly distributed between the two nodes. If I were actually doing this for continued use, I would add
hash $remote_addr;to the upstream so connections from any give IP are tied to a specific node. This means all the traffic from your computer will go to one node, but multiple clients will be load balanced across the nodes.I think load balancing individual requests between nodes is okay, but there might be weird race conditions or different responses that I haven't seen. I would try to tie connections (or at least chunks of requests) to a single node (like done above with the hash of the IP).
@jhollowe commented on GitHub (May 21, 2021):
also note, this will only work when not verifying SSL certificates or updating the certificates on all nodes to allow the IP and/or hostname of the load balancer.