I have installed Fedora Workstation a few times new since Fedora 33 came out.

It defaults to using BTRFS.

But I just installed Fedora 33 Server Minimal from the Netinstall ISO (Fedora-Server-netinst-x86_64-33-1.2.iso) for the first time. and it defaulted to using LVM still.

For my usage of the systems, I do not care which filesystem is in use. I was just surprised to see it different between the two.

All switched and good to go.

Will update clients this weekend. I totally missed that this came down as available last month.

So it is all installed now.

@scottalanmiller said in Raspberry Pi 4 as IT Workstation:

@JaredBusch said in Raspberry Pi 4 as IT Workstation:

This case: https://www.amazon.com/Argon-Raspberry-Aluminum-Heatsink-Supports/dp/B07WP8WC3V
Witth the m.2 add on: https://www.amazon.com/Argon-Raspberry-Board-Expansion-ONLY/dp/B08MHYWJCP

$25 + $20 = a real sexy workstation case.

I need to order that case. I have the 8GB model just waiting to go into a case AND I have a spare M2 to go into it for it!

The second link has an option to get both at once.

hah but not shipping until the 16th because out of stock.

But you can get the case and add on separate immediately.

Samsung, WD, and Crucial are not crap.

@scottalanmiller said in how to push 3rd party software updates to domain clients?:

@Dashrender said in how to push 3rd party software updates to domain clients?:

Chocolately - I use the their package choco-upgrade-all-at-start. This will schedule choco to upgrade at every reboot.

There's another package choco-upgrade-all-at that defaults to 4 AM update daily.

You can just make a Scheduled Task,too.

This does that for you.

@hobbit666 said in PCI bus error:

What OS will you be installing?

Proxmox or raw KVM on Fedora if Proxmox gets too annoying.

I looked at Arctic Wolf in 2019. Was interesting, but not something the client ended up going towards. They did not feel the spend was worth it.

But I thought it was a good solution at a decent price.

@IRJ said in Security Information Event Management (SIEM):

I'm surprised nobody has mentioned elastic yet.

There's an open source version and a free version (more features).

I did not mention it intentionally.

Because it is too complex to use as a SEIM unless you already know a lot about it.

@hobbit666 said in Security Information Event Management (SIEM):

But one minutes it's workign fine then boom errors all over the place

This is not because Elastic is bad, it is because it is complex. Which is why it is a poor solution for people like @JasGot

Unless a person has the time to really learn elastic and how to do things well, it jsut turns into a mess.

So, installing ProxmoxVE 6.3 is simple.

But what is actually the best way to do it?

If it is a single server, you just leave things default, and it installs LVM and away you go.

But what if you want to have 2 servers and use replication?

Well the screw that pooch right off the bat.

As far as I can tell, unless you do everything as ZFS, you cannot do shit.

Has anyone here done anythign with that?

I've reinstalled PVE four or five times no tinkering with things.

@DustinB3403 said in Proxmox VE Setup:

I read through the documentation when I was labbing proxmox, and it specifically calls out that if you need replication the only way to do it is to not use a raid controller (present the drive's directly to the OS).

More specifically, they intentionally restrict it to ZFS. Which really, really sucks. ZFS is a waste of overcomplication.

@scottalanmiller
So WTF is the benefit to PVE over raw KVM for a real business?

Because I am not seeing it. There are exactly zero reasons to ever do anything except hardware raid (enabling blindswap) in any real SMB today.

Without doing ZFS on bare metal, you get almost no beneficial functionality from using Proxmox VE over KVM on your Linux distro of choice.

The real problem with this is the term you used, hosted voip. It does not mean shit technically. So we need to know what you actually have.

It is a marketing term.

@siringo said in Hosted VoIP???:

Could be dependant upon the amount of traffic between sites???

No. SDWAN is the new MPLS, with lower costs.

You get a device for each site and plug it in to your various internet connections. Can be more than one at each site.

It aggregates everything over a set of virtual IP addresses.

So you never "go down" assuming you chose the multiple ISP connections well.

Also your IP never changes. No matter what ISP the traffic is routing on. So things like VoIP don't drop calls when services switch.

But you pay for your bandwith through the aggregator.

It will always be more expensive than setting up your own site to site VPN, but gains you minor benefits. Minor for most businesses. There are absolutely some businesses that are a great fit for SDWAN.

But the thing to remember is that SD-WAN is just running on your existing network connections. so if you only have shit available, the SD-WAN will still be shit.

Also, if you only have expensive lines available, your SD-WAN will be stupid expensive.

So I have played with Proxmox for a couple weeks.

Well the project needs to go live this weekend.

I either need to figure out how to replicate between two servers or switch to a different solution.

The only supported method requires that you not have hardware raid and use ZFS.

That defeats the purpose of blindswap, and shipping drive to the datacenter tech.

Not to mention that ZFS is not even the default setup when the install wizard runs.

@Dashrender said in MPLS alternative:

@scottalanmiller said in MPLS alternative:

We use essentially no files any longer.

I love this -

So you have email
and what Rocket Chat for texting....

Maybe your company has zero need for an individual to make a personal spreadsheet, etc... but if they did, how would you handle that?

That is personal by definition. So they should not be doing it anyway. This is work.

Work "documents" are Excel Online, or WTFever ZoHo is. Not files.

@scottalanmiller said in Promox and VM replication:

The cost of a second server for outages that are generally less than twice a decade and typically quite minor just doesn't justify the cost.

For this client, they can't take a full 24 hours of downtime before they begin to lose data. They need a replica.

It is the far remote sites that send data into the office that drop the data after 24 hours. They were never designed to hold data forever. Machines (pumps actually) with fixed capacities, etc.

There are local printouts that can be manually keyed in later in case of a real large scale disaster at the office location. But an extra $4k for a server that will soon be regionally separated makes it worth it to them.