Do you back-power the RPi with the HDD or do you supply the HDD via the RPi?dropko wrote:I believe this is due to my misuse of the disks attached to the Pi. I have a AC/USB 2.5A capable power supply, which I connect to the Pi (USB). Then my HDD to the other USB port (it powers up the Pi when I attach the HDD, powers down when I detach it).
I believe the latter is causing the issues. Maybe hot-plugging on and off the HDD is not the way to go. Sometimes my TV remote does not (or no longer does) control XBMC and I have to restart it, and I do it by unplugging/plugging back the HDD.
Possibilities are low voltage, overheating, vibrations, shocks (e.g. moving HDD while running) or overextended disks (e.g. running Notebook disks 24/7).dropko wrote:Would too much of this start the disk making those clicks? Could it be power issues?
Make sure file-systems are unmounted properly before detaching power. If XBMC fails to power down, use SSH.dropko wrote:What is the recommended way to turn it off without hurting the hard disks?
It's as above. I supply the HDD via the Pi.Do you back-power the RPi with the HDD or do you supply the HDD via the RPi?
Unplugging HDD or host (RPi) without unmounting the file-systems fries/degrades the file-systems.
The disk stands still as a stone all day. Low voltage I would say is not an issue, but I can't be certain of it. Before my current setup, so said XBMC, I was seeing "disk disconnected / disk connected" all the time during normal play, and that is now gone.Possibilities are low voltage, overheating, vibrations, shocks (e.g. moving HDD while running) or overextended disks (e.g. running Notebook disks 24/7).
Is a halt (shutdown -h) safe for power down? How about powering up, is it ok to just plug it in (like I would do in a PC/notebook)?Make sure file-systems are unmounted properly before detaching power. If XBMC fails to power down, use SSH.
O.K., I wasn't assuming you were, just giving an example of a situation where (unexpected) insufficient (peak) currents gave rise to similar problems. So, in effect you're "back-powering" the Pi. The only issue I have with that is the "boot/power-up sequence is not "well-defined" ie. there's no guarantee that the HDD will "spin-up" etc. before the Pi is ready for it (that's why I power up my USB NAS drive before the NAS box itself which has led to much improved reliability). BTW, have you actually checked that the voltage between testpoints TP1 and TP2 is between 5.25V and 4.74V? It's possible that even a "good" 5V 2.5A PSU with a poor connecting cable may only be providing ~4.75 at the system's peak current. AIUT it's during "spin-up" that the greater current is taken. Once the drive is spinning (at a "constant" speed) the current should be lower & more stable.dropko wrote:@FTrevorGowen
I am not using a Y-cable... My Pi connects to the usb hub directly and my HDD connects directly to the Pi. There is no need for Y'ing, apparently. Just by connecting the HDD to the Pi (directly, no Y cable) makes both power up, and disconnecting powers them down. I bought that "psu" with (advertised) 2.5A in order to avoid any insufficient power issues. That single cabe is enough for power and data, and the disk never spins down indicating power shortage (if that would be such indication).
I have had some success in the past with freezing drives, swapping boards and even taking the top off and starting the platter spinning with a finger. All of these were on drives that would have otherwise be thrown away, and all except the one I changed the board on were only used to recover the data to a new drive, before disposing of the faulty one.elatllat wrote:I have had luck with tapping a drive while it's off to un-stick the bearing and get it to spin up. there are stories of freezing drives, swapping platters (single only) without a clean room and swapping the board on the bottom, but I have never seen them first hand.
I usually use rsync to keep a copy of some systems, or the old dump program to write full or incremental backups to a server or even tape. But that's just what I use. The method isn't important (I've heard good things about rdiff-backup). Getting into the habit of making reliable backups on a regular basis is.Drive failure sucks, rdiff-backup is nice.
So what is the recommended way to do it, supposing you connect directly (USB) and not through LAN?So, in effect you're "back-powering" the Pi. The only issue I have with that is the "boot/power-up sequence is not "well-defined" ie. there's no guarantee that the HDD will "spin-up" etc. before the Pi is ready for it
I haven't. Does this involve a multimeter? That's a bit more technically-detailed that I was expecting to go... I could show the unit I am using but probably such a naming would not be forum-accepted.BTW, have you actually checked that the voltage between testpoints TP1 and TP2 is between 5.25V and 4.74V?
Adding "rootwait" to the line in /boot/cmdline.txt should make the kernel wait until the root device is available before continuing the boot sequence. It's been working for me for almost two years now.dropko wrote:Thanks for all the information! I especially enjoyed the shouting at drives.
@FTrevorGowen:So what is the recommended way to do it, supposing you connect directly (USB) and not through LAN?So, in effect you're "back-powering" the Pi. The only issue I have with that is the "boot/power-up sequence is not "well-defined" ie. there's no guarantee that the HDD will "spin-up" etc. before the Pi is ready for it
I looked up that line. Quite interesting, a useful tip, thanks. I found out also about a "rootdelay", whereas instead of waiting indefinetely (like rootwait), it merely delays it.rpdom wrote:Adding "rootwait" to the line in /boot/cmdline.txt should make the kernel wait until the root device is available before continuing the boot sequence. It's been working for me for almost two years now.
With the on/off switch on the front of my mains-powered HDD caddy.dropko wrote:Still, how do you guys power up/down your Pi+HDD when using direct USB connection?
The on/off switch on this caddy is an electronically-controlled clean shutdown of the power rails, and this particular caddy had previously been in almost daily usage since 2005(?) with various PCs for backup purposes with possibly several thousand non-destructive power cyclings via this on/off switch in that time. Physically yanking a plug out is very different from a clean and controlled power rail shutdown.dropko wrote:It sounds a very hardcore shutdown, to use that switch... just unplugging the usb cord between Pi and HDD (here it seems awfully similar to switch off a caddy or any external powering to the disk).
this is more like my setup.ame wrote:I recently discovered I just needed a beefier PSU to make my Pi + USB hard drive go. Until now I had used two power supplies: one for the Pi (about 1A), and one for the USB HDD caddy (about 2A). I needed this as the 2A PSU was not sufficient to run the hard drive and the Pi at the same time. I modified a USB cable for the hard drive so that the +5V line was cut but the two data lines and ground were left alone. This ensured the hard drive and Pi powers supplies were independent.
Today I tried a 3A USB PSU, since I happened to have one. It's perfect! I can power the Pi from its micro-USB power input socket, and the hard drive takes power directly from the Pi USB port. No splitters/adaptors or headaches.
I'll try it like this for a while longer and see how it behaves. My biggest problem was getting everything started at boot time, but occasionally the hard drive would disappear (then I couldn't replug it). It's a media centre, so it's on all the time, and it's easy to spot when the hard drive decides to stop responding.
So far I have only tried it a couple of times, since the 3A PSU only recently became available.dropko wrote: this is more like my setup.
how do you control your power ups/downs of the Pi?
does your Pi shutdown if you unplug your HDD?
ever remove USB when system is all powered (e.g. to shut it down)?
Firstly there is a big difference between a dead drive("clicking sound of death"/no spin up, smart data fail, etc) and corrupt data.dropko wrote:Are you telling me the Pi is a disk corrupter by nature
what do you mean by this?and if you block the power line from the pi to other USB devices the pi won't be able to dirty the drive power.