choice of external USB drive for mount at boot

Only cut the red wire. Do not cut the black wire as you still need a ground connection. The black wire should be split and connected to the PSU, disk and the Pi.

It's an interesting idea but (for this home server project, anyway --- I'll bear this in mind for later tinkering) I think I'll pay a bit more for reliability instead of splicing stuff together myself!

That's a good point. I've personally come across two kinds of HDD enclosures/adapters: the "full" enclosure (box that you put the SATA/IDE drive inside of, with a USB cable & an external power supply); and the "slide-in" kind. I've only used them for recovering data from drives from machines that had something else wrong with them, & the full-box types I've seen looked quite cheap (the one I bought was in fact quite cheap too).

Wow, how can two models (7200.11 & 7200.14) be so outstandingly worse than all the others --- especially when others from the same brand aren't outstandingly bad?

Sorry if this is a stupid question, but what good does it do me to know (after buying a reputable product) what the physical drive is inside the package?

On the other hand, splicing it yourself could expose a potential unreliability issue - some USB cables have ridiculously thin wires, and this produces voltage fluctuations. Cutting and splicing one would show you the problem, and allow you to replace the cable with a better one.

I guess we should shop for USB cables by weight per unit of length? ;-)

Certainly not. WD Red series is the drive made for 4-bay NAS enclosures marketed to the SMB/SOHO segment. They aren't very fast, but supposedly tuned to not be so bad with wobble from neighbors, or ignore slower SATA bus timeouts that seem to happen more often with those sort of devices.

I don't know what segment of datacenter you'd be in, but WD's "datacenter" drives are the RE's. (ie. RAID Enhanced). I guess they've expanded out recently with Se and Ae as well. Still pretty poor RPMs and throughput for a datacenter drive, I'd say they barely play in that space. (with only on SAS 7200RPM drive available), where commonly we'd use 10k RPM SAS and 15k RPM SAS.

Very easily. Either the technology used to make those models can't stand up to normal operations (ie. can Hitachi Helium drives actually keep their Helium inside and working over a 5-10 year lifespan?), or they were made all in the same plant, that consistantly got bad parts, or some failed process on that line. Ie. a specific plant probably makes all of a specific model, as they have to retool for another model, and if they have multiple plants, they'll spread out their production and save on the cost of retooling and downtime while that happens.

Also, there were rumors that after the flooding in Thailand, that affected many drive makers' plants there, that Seagate mostly just stood theirs up again and kept going with all the stock and tools that were flooded and potentially pitted away or damaged by the sea water.

get

cable to connect data unaltered, and break out the Red (+5VDC) #1 and Black (Ground) #4 wires cut and split so the Pi side is isolated and the drive side is connected to a wall wart mains adapter of sufficient Amperage to prevent back-powering the Pi.

True that, Dom. I had forgotten the device should never ground the shielding, only the host. Oh well, so much for preventing back-powering the Pi AND avoiding a ground loop. So it's back to the powered hub or locating a desktop separately powered hard drive. ?(o=8> wiz.

Interesting point! Anyway, I've decided to go down the desktop drive with external power for this, probably a WD My Book --- unless anyone wants to advise against that.

Interesting, thanks.

I just discovered last night that my powered hub will back power the Pi. We had a nearby lightning strike and everything seemed to have gone wonky. The lights on the Pi were out (other than the power) so I pulled the power cord. The power light didn't go out... I had to pull the hub power cord as well.

I've taking the hub apart, so I knew the hub had nothing to prevent back powering. I guess I just assumed the Pi would prevent that, but once I thought about it I realize with the crude power distribution on the original Pi, it was unlikely. Anyone know if you can back power a B+ or a Pi 2?

You can't boot them by using only back power through the USB. The USB current control circuits prevents it. You need to power via the normal microUSB/GPIO pins/power test points.

However, once the B+/2 has booted up and the USB power circuit has been enabled, the main power source can be removed and the Pi will continue to run on back power.

I think that sucks. There should be circuitry to disable back powering. I would have expected this to be required of the USB hubs as well.

As with everything on the Pi, that comes down to costs - both development and production. Be glad that the newer Pis have better power regulation and at least some USB power control, unlike the originals.

USB hubs aren't supposed to back power. Many do. Again this mainly comes down to costs.

On Thu, 18 Jun 2015 11:16:30 +0100, Adam Funk , in

I have a couple of 2TB WD externals that have held up pretty well -- several years of use. No complaints.

Thanks.

You can then look up the spec/reviews and see if what the makers, reputable or not, have fitted is considered a "decent" or a cheap, slow, unreliable drive.

the

and

...

Why? Unless I'm mistaken the data bus is balanced, bi-directional, half duplex, at least for USB 1 & 2. Just adds another balanced pair to enable full duplex. What ever, no ground is required for a balanced circuit to work.

I'd keep the screen intact which may or may not connect the grouds of each device. Screen to chassis/ground at the host end and floating at the remote end appears to be normal and makes sense.