@scottalanmiller As per the standard, Windows is just ten year old Linux.
Posts made by scottalanmiller
-
RE: Discover What's Inside Windows 11 23H2
-
RE: Discover What's Inside Windows 11 23H2
@Oksana Native 7z? Oh man! It's amazing living here in 2010!
-
RE: Raspberry Pi 5 Announced Today
@JaredBusch said in Raspberry Pi 5 Announced Today:
@scottalanmiller Who cares? Did they address not having the chips to make the 4 in any kind of quantity that makes it available for purchase?
Yes, they moved a ton of their production in house. It's a big deal that they are using more of their own stuff. Won't address everything, but a good move in the right direction.
Also no COVID going on, so hopefully the demand won't be so spiky.
-
RE: Raspberry Pi 5 Announced Today
For those who don't want to read all of the details...
Basically everything from CPU to RAM to GPU is 2x - 3x higher performance than the RP4 from 2018.
Support hats will be available directly from RP for both PoE+ power (no need for external power on these!!) and, drumroll, for M.2 NVMe drives. Meaning you can really use this as a real computer like never before. Lack of native M.2 support was the killer of the RP4 generation.
An RP4 with NVMe would have made it a reasonable desktop system. Without it, it was hard to do anything requiring drive performance of any sort. With more than double the performance across the "board" and NVMe support, this is a very, very powerful workstation.
Active cooling is now semi-standard to accommodate all of this, too.
-
Raspberry Pi 5 Announced Today
https://www.raspberrypi.com/news/introducing-raspberry-pi-5/
Pre-orders already being taken.
Copy pasta of their announcement below...
Today, we’re delighted to announce the launch of Raspberry Pi 5, coming at the end of October. Priced at $60 for the 4GB variant, and $80 for its 8GB sibling (plus your local taxes), virtually every aspect of the platform has been upgraded, delivering a no-compromises user experience. Raspberry Pi 5 comes with new features, it’s over twice as fast as its predecessor, and it’s the first Raspberry Pi computer to feature silicon designed in‑house here in Cambridge, UK.
A Raspberry Pi 5, photographed corner-on, against a plain grey background.Key features include:
2.4GHz quad-core 64-bit Arm Cortex-A76 CPU VideoCore VII GPU, supporting OpenGL ES 3.1, Vulkan 1.2 Dual 4Kp60 HDMI® display output 4Kp60 HEVC decoder Dual-band 802.11ac Wi-Fi® Bluetooth 5.0 / Bluetooth Low Energy (BLE) High-speed microSD card interface with SDR104 mode support 2 × USB 3.0 ports, supporting simultaneous 5Gbps operation 2 × USB 2.0 ports Gigabit Ethernet, with PoE+ support (requires separate PoE+ HAT, coming soon) 2 × 4-lane MIPI camera/display transceivers PCIe 2.0 x1 interface for fast peripherals Raspberry Pi standard 40-pin GPIO header Real-time clock Power button
In a break from recent tradition, we are announcing Raspberry Pi 5 before the product arrives on shelves. Units are available to pre-order today from many of our Approved Reseller partners, and we expect the first units to ship by the end of October.
We’re incredibly grateful to the community of makers and hackers who make Raspberry Pi what it is; you’ve been extraordinarily patient throughout the supply chain issues that have made our work so challenging over the last couple of years. We’d like to thank you: we’re going to ringfence all of the Raspberry Pi 5s we sell until at least the end of the year for single-unit sales to individuals, so you get the first bite of the cherry.
We’re also giving every print subscriber to The MagPi and HackSpace magazines a single-use code, giving them priority access to Raspberry Pi 5 hardware. Click those links to learn more about our Priority Boarding programme — and if you subscribe today, you can get your hands on a Priority Boarding pass too.
Cover of The MagPi magazine issue 143 (October 2023), with "Introducing... Raspberry Pi 5" cover feature
Cover of HackSpace magazine issue 71 (October 2023), with "Eben Upton presents Raspberry Pi 5" cover featureBetween now and the end of October, we’ll be running a series of regular articles and videos, focusing on different aspects of the platform. Keep checking in here.
A little historyWay back in June 2019, we launched Raspberry Pi 4, the first true PC-class Raspberry Pi computer. With a quad-core Arm Cortex-A72 processor clocked at 1.5GHz, it was roughly forty times faster than the original Raspberry Pi model from 2012. In many ways the timing was perfect: in March the following year, schools closed, and millions of schoolchildren around the world were sent to study from home. Tens of thousands of them were able to rely on a Raspberry Pi 4 as their primary PC.
Watch Raspberry Pi 5 show you all of its bits without talkingIn the four years since then, Raspberry Pi 4, and its derivatives Raspberry Pi 400 and Compute Module 4, have become firm favourites of enthusiasts, educators, and professional design engineers worldwide. Modern Raspberry Pi 4 computers run 20% faster than the launch variant, with a core clock speed of 1.8GHz. And, despite the well publicised challenges that have affected the electronics supply chain over the last two years, we’ve made and sold over 14 million units of Raspberry Pi 4 in that time.
But time doesn’t stand still, and neither does our community’s appetite for performance. And since 2016 — the era of Raspberry Pi 3 — we’ve been quietly working on a much more radical overhaul of the Raspberry Pi platform. Today, that effort bears fruit, with the launch of Raspberry Pi 5: compared to Raspberry Pi 4, we have between two and three times the CPU and GPU performance; roughly twice the memory and I/O bandwidth; and for the first time we have Raspberry Pi silicon on a flagship Raspberry Pi device.
New platform, new chipsetThree new chips, each designed specifically for the Raspberry Pi 5 program, come together to deliver a step change in performance.
BCM2712
Close-up photo of part of the Raspberry Pi 5 board, centring the metal shield over the BCM2712 chip, with laser etching identifying the chip.BCM2712 is a new 16-nanometer application processor (AP) from Broadcom, derived from the 28-nanometer BCM2711 AP which powers Raspberry Pi 4, with numerous architectural enhancements. At its heart is a quad-core 64-bit Arm Cortex-A76 processor, clocked at 2.4GHz, with 512KB per-core L2 caches, and a 2MB shared L3 cache. Cortex-A76 is three microarchitectural generations beyond Cortex-A72, and offers both more instructions per clock (IPC) and lower energy per instruction. The combination of a newer core, a higher clock speed, and a smaller process geometry yields a much faster Raspberry Pi, and one that consumes much less power for a given workload.
Our newer, faster CPU is complemented by a newer, faster GPU: Broadcom’s VideoCore VII, developed here in Cambridge, with fully open source Mesa drivers from our friends at Igalia. An updated VideoCore hardware video scaler (HVS) is capable of driving two simultaneous 4Kp60 HDMI displays, up from single 4Kp60 or dual 4Kp30 on Raspberry Pi 4. A 4Kp60 HEVC decoder and a new Image Sensor Pipeline (ISP), both developed at Raspberry Pi, round out the multimedia subsystem. To keep the system supplied with memory bandwidth, we have a 32-bit LPDDR4X SDRAM subsystem, running at 4267MT/s, up from an effective 2000MT/s on Raspberry Pi 4.
RP1Previous Raspberry Pi generations were built on a monolithic AP architecture: while some peripheral functions were provided by an external device (the Via Labs VL805 USB controller and hub on Raspberry Pi 4, and the Microchip LAN951x and LAN7515 USB hub and Ethernet controller chips on earlier products), substantially all of the I/O functions were integrated into the AP itself. Fairly early in the history of Raspberry Pi, we realised that as we migrated the AP to progressively newer process nodes, this approach would eventually become both technically and economically unsustainable.
Close-up photo of part of the Raspberry Pi 5 board, centring the RP1 chip, on which the Raspberry Pi logo and text identifying the chip are printedRaspberry Pi 5, in contrast, is built on a disaggregated chiplet architecture. Here, only the major fast digital functions, the SD card interface (for board layout reasons), and the very fastest interfaces (SDRAM, HDMI, and PCI Express) are provided by the AP. All other I/O functions are offloaded to a separate I/O controller, implemented on an older, cheaper process node, and connected to the AP via PCI Express.
RP1 is our I/O controller for Raspberry Pi 5, designed by the same team at Raspberry Pi that delivered the RP2040 microcontroller, and implemented, like RP2040, on TSMC’s mature 40LP process. It provides two USB 3.0 and two USB 2.0 interfaces; a Gigabit Ethernet controller; two four-lane MIPI transceivers for camera and display; analogue video output; 3.3V general-purpose I/O (GPIO); and the usual collection of GPIO-multiplexed low-speed interfaces (UART, SPI, I2C, I2S, and PWM). A four-lane PCI Express 2.0 interface provides a 16Gb/s link back to BCM2712.
Under development since 2016, RP1 is by a good margin the longest-running, most complex, and (at $15 million) most expensive program we’ve ever undertaken here at Raspberry Pi. It has undergone substantial evolution over the years, as our projected requirements have changed: the C0 step used on Raspberry Pi 5 is the third major revision of the silicon. And while its interfaces differ in fine detail from those of BCM2711, they have been designed to be very similar from a functional perspective, ensuring a high degree of compatibility with earlier Raspberry Pi devices.
DA9091BCM2712 and RP1 are supported by the third new component of the chipset, the Renesas DA9091 “Gilmour” power-management IC (PMIC). This integrates eight separate switch-mode power supplies to generate the various voltages required by the board, including a quad-phase core supply, capable of providing 20 amps of current to power the Cortex-A76 cores and other digital logic in BCM2712.
Close-up photo of part of the Raspberry Pi 5 board, centring the DA9091 power-management IC, on which its name is printedLike BCM2712, DA9091 is the product of a multi-year co-development effort. Working closely with the Renesas team in Edinburgh allowed us to produce a PMIC which is precisely tuned for our needs. And we were able to squeeze in two frequently requested features: a real-time clock (RTC), which can be powered by an external supercapacitor or a rechargeable lithium-manganese cell; and a PC-style power button, supporting hard and soft power-off and power-on events.
Two other elements of the chipset have been retained from Raspberry Pi 4. The Infineon CYW43455 combo chip provides dual-band 802.11ac Wi-Fi and Bluetooth 5.0 with Bluetooth Low-Energy (BLE); while the chip itself is unchanged, it is provided with a dedicated switched power supply rail for lower power consumption, and is connected to BCM2712 by an upgraded SDIO interface which supports DDR50 mode for higher potential throughput. As before, Ethernet connectivity is provided by a Broadcom BCM54213 Gigabit Ethernet PHY; this now sits at a jaunty 45-degree angle, a first for Raspberry Pi, and a source of enduring disappointment for orthogonal-layout enthusiast and CTO (Software) Gordon Hollingworth.
Form-factor evolutionOn the outside, Raspberry Pi 5 closely resembles its predecessors. But, while retaining the overall credit-card-sized footprint, we’ve taken the opportunity to update some elements of the design, to align with the capabilities of the new chipset.
We’ve removed the four-pole composite video and analogue audio jack from the board. Composite video, now generated by RP1, is still available from a pair of 0.1”-spaced pads on the bottom edge of the board.
We now sport a pair of FPC connectors, in the space formerly occupied by the four-pole jack and camera connector. These are four-lane MIPI interfaces, using the same higher-density pinout found on various generations of Compute Module I/O board; and they are bi-directional (transceiver) interfaces, meaning that each one can connect either to a CSI-2 camera or to a DSI display. The space on the left of the board formerly occupied by the display connector now contains a smaller FPC connector which provides a single lane of PCI Express 2.0 connectivity for high-speed peripherals.
Close-up photo of part of the Raspberry Pi 5 board, centring the two FPC connectors, labelled CAM/DISP 0 and CAM/DISP 1The Gigabit Ethernet jack has returned to its classic position in the bottom right corner of the board, after a brief sojourn in the top right on Raspberry Pi 4. And it’s brought with it the four-pin PoE connector, simplifying the board layout at the cost of a compatibility break with our existing PoE and PoE+ HATs.
Finally, we’ve grown a pair of mounting holes for a heatsink, as well as JST connectors for the RTC battery (two pins), Arm debug and UART (three pins), and fan with PWM control and tacho feedback (four pins).
Designed in Cambridge, manufactured in WalesLike all flagship Raspberry Pi products, Raspberry Pi 5 is built at the Sony UK Technology Centre in Pencoed, South Wales. We have been working with Sony since the launch of the first Raspberry Pi computer in 2012, and we’re firm believers in the benefits of manufacturing our products within a few hours’ drive of our engineering design centre in Cambridge: a decade of frequent interaction with the Sony team has helped us understand how to design products that can be built reliably, cheaply, and at massive scale.
Close-up photo of a corner of the Raspberry Pi 5 box, centred on a Welsh flag icon (red dragon rampant on a grey field against a white sky) beside the words "Made in the UK".Raspberry Pi 5 marks the introduction of a number of manufacturing innovations. One of these is intrusive reflow for connectors, which improves the mechanical quality of the product, increases throughput, and eliminates the costly and energy-intensive selective- or wave-solder process from the production flow. Others include fully routed panel singulation for cleaner board edges, and a new approach to production test inspired by our experiences testing our RP2040 microcontroller at scale.
Accessories, accessories, accessoriesEvery new flagship Raspberry Pi product is accompanied by new accessories, and Raspberry Pi 5 is no exception. Layout changes, new interfaces, and much higher peak performance (and a smaller increase in peak power consumption) have led us to redesign some existing accessories, and to develop some entirely new ones.
CaseThe updated case for Raspberry Pi 5, priced at $10, builds on the aesthetic of its Raspberry Pi 4 predecessor, but adds a host of new usability and thermal-management features.
An integrated 2.79 (max) CFM fan, with fluid dynamic bearings selected for low noise and an extended operating lifetime, connects to the four-pin JST connector on Raspberry Pi 5 to provide temperature‑controlled cooling. Air is drawn in through a 360‑degree slot under the lid, blown over a heatsink attached to the BCM2712 AP, and exhausted through connector apertures and vents in the base.
The Raspberry Pi Case for Raspberry Pi 5, showing its red base, the fan assembly with a white frame floating above it, and the white lid floating above thatWe’ve lengthened the case, and tweaked the retention features, to make it possible to insert the Raspberry Pi 5 board without removing the SD card. And by removing the top of the case, it is now possible to stack multiple cases, as well as to mount HATs on top of the fan, using spacers and GPIO header extensions.
Like all our plastic products, the new case is manufactured by our friends at T-Zero, in the West Midlands, UK.
Active CoolerRaspberry Pi 5 has been designed to handle typical client workloads, uncased, with no active cooling. Users who wish to use the board uncased under continuous heavy load, without throttling, have the option of adding a $5 Active Cooler. This attaches to the board via two new mounting holes, and connects to the same four-pin JST connector as the case fan.
The Raspberry Pi Active Cooler mounted on a Raspberry Pi 5. The blower, heatsink, and wires connecting to the Raspberry Pi's four-pin JST connector are visible.A radial blower, again selected for low noise and extended operating lifetime, pushes air through an extruded and milled aluminium heatsink. Both the case and the Active Cooler are able to keep Raspberry Pi 5 well below the thermal throttle point for typical ambient temperatures and worst-case loads. The cooling performance of the Active Cooler is somewhat superior, making it particularly suitable for overclockers.
27W USB-C Power SupplyRaspberry Pi 5 consumes significantly less power, and runs significantly cooler, than Raspberry Pi 4 when running an identical workload. However, the much higher performance ceiling means that for the most intensive workloads, and in particular for pathological “power virus” workloads, peak power consumption increases to around 12W, versus 8W for Raspberry Pi 4.
When using a standard 5V, 3A (15W) USB-C power adapter with Raspberry Pi 5, by default we must limit downstream USB current to 600mA to ensure that we have sufficient margin to support these workloads. This is lower than the 1.2A limit on Raspberry Pi 4, though generally still sufficient to drive mice, keyboards, and other low‑power peripherals.
The white 3-pin UK variant of the new Raspberry Pi 27W USB-C Power Supply, pictured with the cable tightly wrapped with a cable tie and the pins facing towards the viewerFor users who wish to drive high-power peripherals like hard drives and SSDs while retaining margin for peak workloads, we are offering a $12 USB-C power adapter which supports a 5V, 5A (25W) operating mode. If the Raspberry Pi 5 firmware detects this supply, it increases the USB current limit to 1.6A, providing 5W of extra power for downstream USB devices and 5W of extra on-board power budget: a boon for those of you who want to experiment with overclocking your Raspberry Pi 5.
It should be noted that users have the option to override the current limit, specifying the higher value even when using a 3A adapter. In our testing, we have found that in this mode Raspberry Pi 5 functions perfectly well with typical configurations of higher-power USB devices, and all but the most pathological workloads.
Camera and display cablesThe new, higher-density pinout of the MIPI connectors means that an adapter is required to connect our own cameras and displays, and third-party products, to Raspberry Pi 5.
To support existing camera and display owners, we are offering FPC camera and display cables, which convert from the higher-density format (now referred to as “mini”) to the older lower-density format (now referred to as “standard”). These cables are available in 200mm, 300mm, and 500mm lengths, priced at $1, $2, and $3 respectively.
Two orange cables crossed at one far end lying flat on a plain grey background. White writing on the each cable says "Raspberry Pi Display Cable Standard Mini 200mm" with a Raspberry Pi logo in white and other legal safety logos in whiteCamera Module 3, the High-Quality Camera, the Global Shutter Camera, and the Touchscreen Display will all ship with both a standard-to-standard and a 200mm mini-to-standard cable.
PoE+ HATFrom early 2024, we will be offering a new PoE+ HAT. This supports the new location for the four-pin PoE header, and has an L-shaped form factor which allows it to sit inside the Raspberry Pi 5 case without interfering mechanically or disrupting airflow.
A visibly hand-soldered prototype of the L-shaped Raspberry Pi PoE+ HAT for Raspberry Pi 5Prototype PoE+ HAT. We don’t know yet what the production version will look like, but we do know that it won’t look like this.The new PoE+ HAT integrates a planar transformer into the PCB layout, and utilises an optimised flyback converter architecture to sustain high efficiency across the whole zero to 25W range of output powers.
M.2 HATsOne of the most exciting additions to the Raspberry Pi 5 feature set is the single-lane PCI Express 2.0 interface. Intended to support fast peripherals, it is exposed on a 16-pin, 0.5mm pitch FPC connector on the left-hand side of the board.
From early 2024, we will be offering a pair of mechanical adapter boards which convert between this connector and a subset of the M.2 standard, allowing users to attach NVMe SSDs and other M.2-format accessories. The first, which conforms to the standard HAT form factor, is intended for mounting larger devices. The second, which shares the L-shaped form factor of the new PoE+ HAT, supports mounting 2230- and 2242-format devices inside the Raspberry Pi 5 case.
Prototype of the larger (standard HAT form factor) M.2 HAT, mounted on a Raspberry Pi 5Prototype M.2 HAT. Final hardware will not look like this.
Raspberry Pi Beginner’s Guide, 5th EditionSporting a brand-new look and feel, and priced at RRP £19.99 ($24.99), this new edition of our bestselling Raspberry Pi Beginner’s Guide is the definitive manual for Raspberry Pi computers and accessories. It has been comprehensively updated to cover Raspberry Pi 5, and the upcoming release of Raspberry Pi OS based on Debian Bookworm.
RTC battery
RTC coin cell connected by red and black jumper wires to a two-pin JST plugLast, but very much not least, we have sourced a Panasonic lithium manganese rechargeable coin cell, with a pre-fitted two-pin JST plug and an adhesive mounting pad. This is priced at $5, and is suitable for powering the Raspberry Pi 5 real-time clock (RTC) when the main power supply is disconnected.
A newer, better Raspberry Pi OSIn parallel with the final stages of the Raspberry Pi 5 programme, our software team has been busy developing a new version of Raspberry Pi OS, the official first-party operating system for Raspberry Pi devices. This is based on the most recent release of Debian (and its derivative Raspbian), codenamed “Bookworm”, and incorporates numerous enhancements, notably the transition from X11 to the Wayfire Wayland compositor on Raspberry Pi 4 and 5.
Raspberry Pi OS will launch in mid-October, and will be the sole supported first-party operating system for Raspberry Pi 5. Keep checking back here: we’ll be telling you some more about the new OS, and you’ll be able to download it shortly before Raspberry Pi 5 arrives on the shelves in late October.
CreditsBringing Raspberry Pi 5 to life has been a seven-year, $25 million endeavour, involving tens of organisations and hundreds of individuals. A non-exhaustive list of those who have contributed to Raspberry Pi 5, and its constituent silicon programs, is here.
-
RE: What Are You Doing Right Now
@Obsolesce said in What Are You Doing Right Now:
@scottalanmiller said in What Are You Doing Right Now:
Kept them up for 2.5 years.
Tf?
Right? One reboot and.... ZFS can't find any drives because ZFS uses ephemeral designations for its drives and has no known way to discover drives once that ephemeral identification is lost. So any random reboot can cause the array to just... vanish, even though ZFS can see all the drives just fine and knows that they are ZFS drives.
-
RE: ProxMox Storage Configuration Question (idk how lol)
@JaredBusch said in ProxMox Storage Configuration Question (idk how lol):
@scottalanmiller said in ProxMox Storage Configuration Question (idk how lol):
It means some random drive delivery guy can do the drive swap without asking.
It means the "dell" or "hp" tech that is just a random contractor can swap the drive under the warranty support wihtout needing us to deal with anything more than checking that the autorebuild started in the controller.
Right. Not something I recommend doing, ever. Because that's the same people who pull the wrong drive or the right drive from teh wrong server. It's a great idea, and that's why we normally do it for really small shops. But it carries a lot of dangers of its own because it encourages people to make big hardware changes without asking.
Seen a LOT of data loss causes by making this seem like a good idea.
-
RE: ProxMox Storage Configuration Question (idk how lol)
@JaredBusch said in ProxMox Storage Configuration Question (idk how lol):
@scottalanmiller said in ProxMox Storage Configuration Question (idk how lol):
In general, I actually think this is a negative. Making systems that are "easy for people who don't know what they are doing to pretend that they do" is one of the biggest causes of problems I find in customer systems.
No one knows how to use MD or ZFS. Instead they go to Google-sensei.
End users, sure. Sales people, sure. But if you have an IT team, you are good to go. Having to pay hundreds of dollars for lower reliability, lower performance solutions to allow shops without IT to pretend to keep themselves safe is a penalty for people who don't want skilled labor. But overall, just hiring an IT team or having a qualified IT department makes far more sense. You get more protection and often actually costs less. There's no shortage of IT people.
I'm a huge believer in doing a good job and if the customer screws up not caring, that's on them. But intentionally doing a bad just assuming the customer is an idiot makes it my fault.
-
RE: ProxMox Storage Configuration Question (idk how lol)
@JaredBusch said in ProxMox Storage Configuration Question (idk how lol):
I use it because it is more easily understood and supported by the majority.
In general, I actually think this is a negative. Making systems that are "easy for people who don't know what they are doing to pretend that they do" is one of the biggest causes of problems I find in customer systems.
"Oh, I thought I could just change how these things work and..." now they have no backups, no they are offline, now their data is corrupt, etc. etc.
The appearance of being accessible without knowledge encourages the Jurassic Park Effect.
-
RE: ProxMox Storage Configuration Question (idk how lol)
@JaredBusch said in ProxMox Storage Configuration Question (idk how lol):
I don't use it (hardware raid) because it is better, I use it because it is more easily understood and supported by the majority.
When non-IT people need to interact, it's better. That's why we use it. We don't want customers thinking that they can do stuff without IT and causing damage. It means we can send a middle schooler in to change drives without needing to coordinate on the timing. It means some random drive delivery guy can do the drive swap without asking.
It costs a lot more. It lowers performance. But the blind swap value for customers without someone technical making sure random people aren't touching servers when they aren't supposed to is a big deal.
However, all those non-technical people are still hitting the power button, pulling cables, spilling coffee... so I don't know if it has ever protected us, lol.
-
RE: ProxMox Storage Configuration Question (idk how lol)
@JaredBusch said in ProxMox Storage Configuration Question (idk how lol):
Software raid is bespoke.
That's true in the case of ProxMox with ZFS, because ZFS isn't native nor stable on that platform nor is there any team that makes that work. It's a product from one world shoehorned to allowed it to run (possibly in license violation) in another and there is no official support, testing, or anything from any vendor or team.
But MD is way, way the opposite. MD is as enterprise and anti-bespoke as it gets. It's part of the base OS, it is the most reliable RAID system out there.
Technically, using a third party external hardware replacement for the OS' own tooling is far more bespoke than using MD internally.
-
RE: What Are You Doing Right Now
@Obsolesce said in What Are You Doing Right Now:
@scottalanmiller said in What Are You Doing Right Now:
Also, today is yet another day of slowly and expensively helping a company recover both their ProxMox and TrueNAS deployments that were done on ZFS and all was lost and no vendor had any means of recovering anything.
Did they back up their PRoxMox to their TruNAS and they both went under?
no, no backups. Data was all stored on TrueNAS. TrueNAS was stored on ProxMox. "ZFS is so reliable you don't need backups" was the idea, I guess. And no hardware failed. Just ZFS failed. Pure software failure. All hardware is pristine and working great. But ZFS just lost... everything due to internal design fragility that makes it prone to data loss on reboot.
Original IT company seems to have known this and disabled all reboots. Kept them up for 2.5 years. First reboot after going live resulted in total data loss because ZFS on Linux isn't expected to reliably survive reboots in that way!
-
RE: What Are You Doing Right Now
@travisdh1 said in What Are You Doing Right Now:
@Obsolesce said in What Are You Doing Right Now:
@scottalanmiller said in What Are You Doing Right Now:
Also, today is yet another day of slowly and expensively helping a company recover both their ProxMox and TrueNAS deployments that were done on ZFS and all was lost and no vendor had any means of recovering anything.
Did they back up their PRoxMox to their TruNAS and they both went under?
Or the ghost story of IT deployments.
ProxMox with main storage on TruNAS... ooooooOOOOOOo scary!
NO! Worse. TrueNAS on top of ProxMox. So ZFS on top of ZFS!!
All the ZFS cultists are so ready to say why you can't do ZFS on top of hardware RAID to try to discredit hardware RAID (when in reality, it is ZFS that is unstable) that they ignore that doing it on top of ANY other system would have the exact same results. So they miss that ZFS on top of ZFS is far, far worse than ZFS on top of something stable.
So the level of disaster is incredible. All teh complexity, all the fragility of ZFS (TWICE) and no ZFS advantages because it is stripped away by the encapsulation of the ZFS by the other ZFS system.
-
macOS 14 Sonoma
macOS Sonoma 14 has released this morning. Mac users can update now. I'm just about to kick off the update myself.
https://www.apple.com/macos/sonoma/
No major features that I would call important in any way. but lots of little things.
-
RE: What Are You Doing Right Now
Also, today is yet another day of slowly and expensively helping a company recover both their ProxMox and TrueNAS deployments that were done on ZFS and all was lost and no vendor had any means of recovering anything.
-
RE: What Are You Doing Right Now
@GUIn00b said in What Are You Doing Right Now:
@scottalanmiller said in What Are You Doing Right Now:
hating ZFS more and more each day, lol
I've quickly discovered there is a fervent populus of ZFS defenders and apologists. It reminds me of the RAID-5 knife fights ....err I mean discussions
You sure do know how to pick a nemesis, SAM
More they picked one with me. I've been a storage and filesystem expert since before ZFS released and the ZFS team worked on the original SAM-SD design with me. The Cult of ZFS people literally picked a fight with me and the ZFS team ten years after this stuff was old hat, lol.
-
RE: ProxMox Storage Configuration Question (idk how lol)
@travisdh1 said in ProxMox Storage Configuration Question (idk how lol):
They've completely bought into the cult of ZFS and the Windows world of "software RAID is bad".
It's weird because they push software RAID, just not good software RAID that's baked in. Only software RAID from a system that isn't native nor stable on Linux. Ugh.
-
RE: ProxMox Storage Configuration Question (idk how lol)
@GUIn00b said in ProxMox Storage Configuration Question (idk how lol):
I thought I was going to mdadm a RAID-10. Nope! AAaaaand Proxmox seems to have "ZOMG USE ZFS" all over their documentation, and I'm not interested in that if there are other options
The only way I'd use it is either manually creating a RAID array with MD and not telling ProxMox, lol. But that's kludgy. Or do things the ProxMox way and use a hardware RAID controller.
ProxMox isn't stable with ZFS (ZFS is not stable on Linux and we know exactly why) and ProxMox ignores this and leaves the user at risk. But almost no one deploys Proxmox that way so it is rarely an issue.
But like VMware, officially there is no software RAID option that is production capable. Sucks because MD would do a great job here.
-
RE: Practical RAID Decision Making
@GUIn00b said in Practical RAID Decision Making:
I guess it would have to be a very specific concern to opt for the parity overhead in favor of the "added protection" over a statistically very rare potential failure scenario of 4-drive RAID-10.
It's a specific failure scenario that even when it happens, there's no way to know if the same scenario would have been protected under RAID 6 because most scenarios where RAID 10 would fail, RAID 6 would also fail during its recovery mode (nearly 100%.) But the chances that it would face that recovery scenario are higher.
The complexity comes from choosing single unpredictable failure scenarios. After a failure has occurred, if we had the ability to pick how to have protected against it in the past, yes, RAID 6 would be chosen sometimes. There's a known example to explain why you can't use this in real life. It's the seatbelt problem.
Seatbelts save lives. On average, by far, wearing a seatbelt protects you. But there are special cases where the seatbelt can be what causes you to die. Yet statically, you never skip wearing a seatbelt because it is a one in a million chance that the seatbelt will cause a death rather than preventing one. And at the time that you choose to wear or not to sear wear your seatbelt you have no idea which type of accident you will have.
So we know that wearing the seatbelt is the safer bet. Seatbelts are like RAID 10. You can't know how things will go wrong, and in this scenario, RAID 10 protects you much more often than RAID 6 does.
-
RE: Practical RAID Decision Making
@GUIn00b said in Practical RAID Decision Making:
Though both RAID levels can sustain 2 drive failures, the caveat with RAID-10 is as long as it's not the same member from each mirrored set. With RAID-6, ANY 2 drives could fail and still be operational and recoverable. I guess it would have to be a very specific concern to opt for the parity overhead in favor of the "added protection" over a statistically very rare potential failure scenario of 4-drive RAID-10.
Even then, statistically RAID 6 is much more dangerous. RAID 10 has a reliability rating so high that it never matters, RAID 6 does not. RAID 6 has a rebuild time hundreds of times longer than RAID 10; and it has 300% higher URE risks during a rebuild (that is chance of hitting one in a four drive scenario).
Remember the rule of thumb in determining RAID risk: always ignore the false security of "how many drives can you lose." That's not what matters. That's one of many factors, and almost never a significant one, in determining actual risk. URE risks are orders of magnitude more significant and factors like rebuild intensity and rebuild time make "chances to lose another disk" generally more significant than "how many disks can you stand to lose."
At four drives, I know of no scenario where RAID 6 is faster or more reliable than RAID 10. It's always worse. At 5+ drives it starts to have capacity advantages that once in a while make it a good choice. But the rule is at four drives, RAID 6 is a "never" because it's slower and riskier without any offsetting benefits.