Disney is raking its customers over the coals with a 75% price hike for their annual subscription (originally $80.) People wonder why piracy is on the rise.
Disney is raking its customers over the coals with a 75% price hike for their annual subscription (originally $80.) People wonder why piracy is on the rise.
deleted by creator
ECC RAM is only necessary for people doing financial-related work.
If a video has a bitflip that is not corrected in software, ooooo 1 pixel will be a slightly different shade or hue or one subtitle letter will be wrong worst case.
Billing, payment processing, virtual currency storage, a flipped bit could be thousands of dollars, but those systems will have multiple verifiable redundancies in place, unlike the 90s when people like to quote that ECC RAM is essential.
Also 100% uptime servers like enterprise storage servers where customer data integrity is high priority.
I have yet to see a single shred of evidence that a memory bit flipping has caused any problems past 2008 or so. Maybe another person has found some case where it has, but when I was researching for my own server, I couldn’t find a single one.
Nearly every problem (1 million times more likely) is caused by software instability and bugs, with some being due to hard drive bit rot or hardware failures which ECC won’t fix anyway.
Not server-related, but an instance where an inexplicable bit flip caused a stir is Super Mario 64 speedrunning. There is a level that is notoriously slow to navigate and during a playthrough a community member “discovered” a skip that warps you about halfway through the level. There is a video of it happening on live stream, but to this day someone has yet to reproduce the skip. Fiddling around with the game’s memory showed that the behavior happens when a single bit is flipped. All in all, it was likely a one-off error on the hardware that happened at exactly the right time in exactly the right place. The incident is known as the “TTC upwarp” and there is a $1000 bounty to claim if you can provide a working set of instructions to reproduce it on real hardware.
I mean, that was actually pretty cool to read about! Speedrunning community always does the most crazy things as far as hardware memory dumping and analyzing to drop time in a speed run. 😅 that is passion.
It did happen on a device from 1996 though where in the time, programming and error checking was so barebones and efficient that a single bit could really mess a lot of things up.
That’s why I specified a time period 😉. Originally bug were called bugs because literal bugs would get in the holes of punchcards and make programs not run. Not a problem anymore! In the same way, systems have implemented checksums and error checking such that it really isn’t a big deal for the vast majority of applications.
To be completely honest, I kinda did an oopsie because it completely slipped my mind that although it happened in 2020, the technology involved is indeed pre-millenium.
It’s before 2008, but a bit flip changed a Belgian election.
This is part of the reason I keep my servers in my basement.
For Plex? No. It’s running on a $150 Amazon minipc lol
What does that 50TB look like? I’m pushing up against 25 at this point but it’s shamefully all usb hdds plugged into a usb strip in a mini pc, and it’s less than ideal.
That’s filthy
Buy a NAS unit such as a Synology DS923+
Add 4 drives to it of equal size. One drive’s worth of space will be sacrificed for redundancy and they’ll all be combined into a single storage drive.
I have 4x 18TB drives giving me just under 50TB of usable storage. Any single drive can fail with no data loss, and I just replace it and keep going.
Removed by mod
Good point. I’m personally not doing any transcoding so it doesn’t affect me.
Removed by mod
I’m using a mini PC for the transcode so the proc in the nas doesn’t matter, it’s just storage. The mini pcs on Amazon are way more powerful than anything you’ll find in a standalone
I ripped out all my portable drives, had to make a cable without the energy management line for inside a PC.
Big ATX case with 8 3.5’ bays, mobo with 6 Sata ports, last two ports will require a SATA-PCIe expansion card.
1x 4TB (shucked)
2x 6TB (shucked)
1x 8TB HDD
1x 16TB Ironwolf Pro
40TB across 5 bays, 1 left without expansion card, 2 more with the card, following that the 4TB will get dropped for a bigger drive.
I’ll likely be buying no less than 12TB per drive going forward, no RAID configuration yet.
I love it. That works fine but if you want a weekend project consider doing something like this:
Buy a Fractal Define 7XL.
Shuck all those HDDs and put them in the case. Buy one more disk, as large or larger than the largest disk.
Buy 8GB RAM, cheap mobo, cheap Pentium gold CPU, and reasonably reliable power supply. Also buy either a SATA or HBA PCI adapter.
Install unRAID and Plex. Use that extra drive for parity. Now your data is protected if a drive fails. They also won’t get so hot, and speed will be much better. You also have lots of capacity for more drives.
deleted by creator
Amazon has Chinese usb enclosures for $150, that’s what I’m using.
deleted by creator
Fuck no
When I built my latest Plex server, I chose to put ECC RAM into it. But it was a pain getting all the hardware, due to the silly rules AMD has for ECC support and iGPU support in its chips.
What? Unless you’re running an actual commercial scale website, no, nobody needs ECC.
If you are planning to use zfs its worth a few extra bucks for ecc.
It can be far more than “a few extra bucks.” It’s not just buying more expensive ram, but you need a compatible motherboard and processor as well.
It can be. It doesn’t have to be. If ECC is too expensive I recommend avoiding ZFS.
Like ECC there’s little to no benefit for the average user to use ZFS.
I agree. I didn’t realize we were talking about average users. I thought we were talking about people building NAS.
I’m talking about the average person building a media server to cut out monthly streaming or cable costs. They don’t need or even know ZFS.
That’s why I agreed with you. :P