On Thu, May 29, 2014 at 07:12:24AM -0500, Johnny Hughes wrote:
On 05/28/2014 10:52 PM, Fred Smith wrote:
I didn't do that, but what I did do was download a nightly build for Fedora 21, a live CD image, burn it to a disc and boot that. One would think it has a recent kernel (I neglected to look while it was running to find out what kernel it was).
so I ran a bunch of commands to copy stuff to and from it, and while it took a few minutes, it also failed in more or less the same way. I'd guess that means that whatever the bug is, it's not fixed in Linux, yet.
:( :( :(
Johnny:
I believe that is the only USB3 device here. Did you see my posting a few days ago that said it appears to work as a USB2 device? At least I beat on it for a good while and it didn't fail, then.
So, do you have any other external USB 3 drives laying around to test with?
It would be good to test if this is a problem with more than just the one external RAID box that you are using.
How long does this usually take to get the error and what kind of writes access does it take to cause it to fail .. I have 2 of those exact boards in production and a USB3 drive I can plug in and test. I would like to know if this is specific to individual drives or an overall USB3 problem as I recommend those boards to people now.
(yeah, I know... you may recall I asked about good boards and you said you'd had good luck with that one, so in lieu of other suggestions I got one. So far it's been great!)
What I've done to make it fail is find a file (or a tree of directories and files) of several gigabytes and copy it to the device repeatedly until it fails, removing the file from the device between copies. It usually takes only 2-3 such copies to see the failure.
FYI, the Fedora nightly build live CD (the Mate one) I ran last night also suffered a kernel error right after the USB failure (I suspect they may be related, but don't know with any certainty) and auto-generated a bugzilla report, if you think it may be interesting, it's number 1102452 which has, as I now see, been closed as a duplicate of 1096572..
Fred