What kind of headaches are you having? I’ve been running two completely different machines in a cluster with a pi as a Qdevice to keep quorum and it’s been incredibly stable for years.
What kind of headaches are you having? I’ve been running two completely different machines in a cluster with a pi as a Qdevice to keep quorum and it’s been incredibly stable for years.
Not if you’re serving content to a device that can do the decoding, like a Shield. My Jellyfin server runs in a Proxmox VM with no GPU passed though, and transcoding disabled for all accounts.
Gitlab is pretty great!
Usenet and hard drives. LOTS of hard drives.
Ooh, I know this one! Here’s what I did to get it working and set so it survives a reboot:
sudo cp ~/.config/monitors.xml /var/lib/gdm3/.config/
Note: You may need to do this as sudo -i
Servers and computers get Ankh-Morpork street names.
The robot vacuum cleaner is GLaDOS.
SANGUIS DEO SANGUINEO
That is crazy. According to a comment on that article, most BIOS uses UTC (as does Linux, obviously), but Windows uses localtime for some reason, so it converts UTC to localtime after boot, then back to UTC when it needs to do little things like networking or TLS.
If you’re rural, check out B4RN.
Hell, I can get a 30 year old HP LaserJet 4 printer working just fine on almost any version of Linux with the official HPLIP CLI software provided by (shockingly) HP, which was updated 2 months ago with support for over 50 new printers and the following OSes:
I HATE HP and their printers (PC LOAD LETTER WTF FOR LIFE) but I will admit that this is impressive support.
And Windows is used on business PCs largely because of how manageable they are at scale.
… Linux being manageable at scale is kind of the reason why Linux is the standard for servers. Many enterprises run Linux workstation distros, and they can be managed at scale just fine, it’s just different tooling. You can deploy a Linux desktop OS with Ansible as easily as a Linux server.
You can replace pretty much the entire Office suite with Nextcloud and OnlyOffice, both of which can be easily hosted on-prem, for a fraction of the cost of paying MS for roughly the same thing on their awful infrastructure.
If it was feasible for business to change to a free alternative, I guarantee they would’ve done so.
They have. Just because you haven’t heard about it doesn’t mean it isn’t happening. It’s pretty easy (and inexpensive) these days to run Linux desktop OSes like RHEL, Debian or Ubuntu on a VM running on Proxmox or OpenShift, complete with multiple monitor support and GPU. Hell, you can even run a Windows VM if you want. All you need is a system (like a thin client) with enough grunt to run a browser, and enough ports to handle multiple monitors and USB accessories.
And businesses aren’t interested in “free”, they’re interested in support, which they are willing to pay for. This is how companies like Ubuntu, Red Hat and SUSE make their money. The OS is free, but you can pay for professional support.
Unfortunately not, it applies to all clusters. A quorum requires at least 2 votes, which means 2 nodes. But if one node goes down, you only have 1 vote, and the cluster will go into read-only mode, which means you’ll lose the GUI and the ability to manage your nodes:
When your cluster is non-quorate (so at least half of all nodes are dead), the remaining nodes will change the PVE management into read-only mode. Because of that you will no longer be able to change and manage your VMs and containers and will also not be able to log into the GUI. This is done to avoid cluster split-brain problems in which they run into inconsistent states. (Source)
But if you have a device that will supply a vote in the event that one of the two votes is unavailable, the cluster will continue to function with a single node, which will allow you to use the normal Proxmox tools and interfaces to diagnose the problem, while also keeping any VMs on the single remaining node up and running (available).
Here is a Proxmox employee explaining it a bit more clearly than the official documentation:
… a cluster needs to be always quorate to work properly, not just for HA. High availability just means that the cluster will try to keep your HA-enabled VMs and containers always available, i.e. if a cluster node fails the HA-manager will launch HA-managed guests on another cluster node.
While a 2 node cluster should work with 2 active nodes, if one of your nodes goes down your cluster will automatically be non-quorate and will no longer work as expected. To have quorum in your cluster, you need a setup of at least 3 nodes, though you do not need 3 full Proxmox installations … you can setup something like a Raspberry Pi as a QDevice for external vote support.
(Source)
You may have trouble keeping a quorum with a proxmox cluster of 2. You should really have 3, or set up a q-device. (Source)
Each node in a cluster has a “vote” in a healthy cluster. If one of your devices goes (or is taken) down, you’ll lose quorum, and the GUI and some other stuff. You need 3 votes for things to work reliably. I have an old RPi set up as a q-device and it works fine.
That sounds like the business model of Substack
Tubesync is pretty great