risk

Look at your inventory, send them an email and introduce yourself and say how many devices you manage and say you're new to procurement and you'd like some help coming up with a proposal to steer your organization towards more sustainable Framework, and you'd appreciate if they can spare a short 20-30 minutes introductory call to maybe help you out and help you structure the pitch. (beware Dell, Lenovo, HP sales reps are brutally empowered to slash prices like there's no tomorrow, pitching them against each other works well if you have enough budget to throw around, Framework might not have the scale to do this).

My guess is the 750G/day limit is there to prevent ddos type attacks, where you'd create a bunch of accounts over time and sync your botnet to try to upload petabytes of random per second burning through the accounts as it goes through. Backups of stuff sound like a legit use case, and you're still paying a hefty chunk, I don't know why anyone would be mad. On the contrary, it's actually a creative way to use the service, I wish you did a segment where you explain how folks should setup rclone, but maybe that's more for wendell

> DHCP DHCP is a service that needs to exist on these VLANs, a "server" is a piece of software that provides a service. A machine is a piece of hardware, .. there's hosts (can be desktops or servers or laptops), and then there's routers and switches etc etc... DHCP is extremely light weight, you don't want multiple machines just for DHCP, you may want multiple server grade machines connected using trunks to all VLANs and then you can migrate DHCP running in VMs across them as needed (e.g. during upgrades or hardware failures). > DNS pretty much all decent DNS servers support DDNS and split horizon, a single VM sitting somewhere for DNS is fine, a single IP is fine. > file server parts.. you need a single ip, should secure access to files based on accounts primarily, and then optionally additionally based on network (as your second layer of just in case security), samba allows you to configure which shares and users have access based on ip they're coming from, same with printers. > *.abc.local with devices roaming across networks and the whole byod world, a better practice is to register a real domain, and use client and server certs, so your device works.

Yes. Blocks more than rads, you could run 2x420watercool mora radiators and it'd be fine, but 7 GPUs and 2 CPUs should get you thinking.

If your router supports OpenWRT/LEDE you could use it split off a second VLAN for servers. What router model do you have? If not, you could either get a Ubiquiti edgerouter-x or a Mikrotik HEX, they both support VLANs and a whole bunch of other features for $60, and you'd then turn your current router into an accesspoint.

Sketchup is cool, you may want to give Fusion 360 a go as well, it's less dragging and pulling and punching in numbers, but it supports parametric design, which is considered a really nice feature to have in the industry. Parametric design is that thing where you add constraints like saying you want this line parallel to this other line, or that dimension parallel to that other dimension, so that when you resize things later your model stays together where it should and stretches where it should. I mention it because it looks like you look motivated to do CAD at the moment - might as well use the opportunity.

It sounds weird that it's thermal throttling, have you tried reseating the GPU block? Maybe someone lurking from EK can comment

One of the great things with opensource software is that if you have the will and skill you can fix things and improve on them. More importantly you can easily learn from it, educate yourself, and eventually have a greater chance of contributing to the sum of human knowledge. That said, another way to contribute to the sum of all human knowledge is to develop something useful, sell it to consumers, invest into research and accumulate value and accumulate sales. Usually, lectures in universities are free for anyone to attend. Most companies implement some in-house training, or do knowledge-transfer and sharing between employees in some form. You have your equalizers in opensource and universities and your diversity / imbalance creation in companies and economies - those two forces will always tug on each other, they can't exist without one another, it's one of the ways in which this civilization works. That said, I think if you're forcing people to go out of their way to fix or get support your product, it's a shit product. As a user/consumer you should absolutely have the right to go elsewhere for advice and support, I think that right should be regulated and protected and anyone should be able to offer support and repair for the product. The bit on repair.org that says "It requires manufacturers to provide owners and independent repair businesses with fair access to service information and replacement parts" I think that makes sense for the company manufacturing the product as well. The company making the product already has the advantage so if the business is managed well, the risk to business not only that it's not high, but it actually increases the value of the product and de-risks how to handle issues. For example, if you need to do a recall, it's much cheaper to have all that workforce of repairmen at the ready - having it decreases business risk.

is that air or something else, the bubbles I mean, is your pump reservoir pulling air and vortexing ?

Define C is smaller, I'm thinking Define S is a better choice as it'll will leave you with more space to properly lay out your water cooling components and would give you less stress when filling / refilling your loop (also, dual 360 maybe?) Also, x360 is pretty much the same price as p360

which wifi card / driver are you using (or trying to use)? is there anything related to wifi in journalctl -b ?

Was this measured on Ryzen? I'm asking because on those CPUs the "Infinity Fabric" connecting CCXs runs at ram speed, so as a hack to increase that speed, people try to get faster ram.

Are you seeing any instability as a result? a friend was running a 5930k cooled by an x62 at high temps - trying to optimize for noise.... Turns out that the CPU didn't respond well to clocks changes and was unstable, once in the 70s it was fine.

tried vlc?

Seconded.

>... server boards ... that xeon should work on a consumer h81 board .. as Droidbot said.

If you have any backups or use cloud services, they can probably get those. Here's a interesting tidbit, when someone steals a laptop to get stuff off of it, they won't shut it down. It'll still be powered on and will have the keys in ram, preventing syphoning of those is important as well. The contents of the ram and keys can be grabbed while the ram is powered on and rest of the machine is asleep, I'm not sure if Ubuntu by default locks your volumes before sleeping - if not ... you need a couple of things.... Automatic updates - set to download but not install. If your machine is older and still has FireWire, disable it, if you have thunderbolt and no vt-d either on chipset or cpu, look into disabling thunderbolt and using it as a displayport only. If you have vt-d look into securing thunderbolt using it. There's also some USB guard software that Ubuntu doesn't have by default, normally USB drivers load and the device and driver are trusted by the machine. Since the machine wakes up from sleep, and it's running and has a network port probably, you need a firewall. Make sure you have a firewall setup to drop all incoming packets except ones belonging to previously established outgoing connections, and ones related to servers you intentionally want to run. Drop all outgoing packets unless the source port is one of those ephemeral source port ones, or if they're related to connections to existing servers. Drop traffic on 'lo', unless it's for a server you intend to run. That will make it harder to chain exploits, by not allowing an exploited app to bypass the firewall. Commonly I see people just accepting any traffic on lo - makes me sad... if you're a dev and need to test with localhost, just make a veth and put it in a namespace, don't use the public host networking stack.

in terms of cash, sell the 390s and get a 1080ti..... it'll run cooler, quieter and faster for same money, then get water cooling.

You can try Antergos It's basically Archlinux for noobs / lazy folk.

LVM is a Logical Volume Manager It enables you to do snapshots and thin provisioning and you can have as many partitions as you want. (kind of like with ZFS and btrfs), but it does this on a block level. You can go back to windows and shrink the windows partition, it might take a few tries to get enough space and enabling/disabling some settings that cause windows to keep data close to end of the partition. (I've done this last year on my mom's computer when switching her to an SSD, don't remember the details, sorry).

assuming you're using google docs, you can either "insert > equation" or you can "format > superscript".

tar xavf myarchive.tar.gz what does it report?

how much is a dell xps 15 with 32gb ram there?

I think you can get by better if you were making stuff in sufficient volume, steel isn't that expensive. In volume (e.g. 500/1000 pieces), there's cases you can get for $5 a piece off of Alibaba ... the scary bit is you can get supposedly 400+W power supplies for $5 too. Let's say you can fit 1000 cases into a 40' container -- costs about $2000 to ship ($2 / case) ... but wait, there's good news - and that's that computer cases are mostly empty -- you can stuff them with ram iphone cases, adapter cables, RGB strips, and other fluff... (imagine stuffing it full of RGB LEDs and then keeping all of them - plugging them in

You can probably delete, or mv that file out of the way.