Re: disk error -> reset entire USB connection
The drives/partitions are all identified by UUID or NAME, not
/dev/sd*. The external disks all have external power.
I should have been more specific about my main question: given a
failure on one drive, is it to be expected that the connections of all
the drives sharing the same USB connection will reset?
But even if the connection is reestablished the disruption is
obviously not good for I/O taking place at the time of the
interruption, and it's not clear to me that things will work OK
afterwards.
The lvm-raid wiki says not to use RAID over a USB connection. I ran
into the same problem with the replacement drive, and it (so far) went
away when I moved it into the main computer case.
Ross
On Fri, Feb 3, 2017 at 6:59 AM, Mark Fletcher <mark27q1@gmail.com> wrote:
> On Thu, Feb 02, 2017 at 06:00:59PM -0800, Ross Boylan wrote:
>> I have multiple drives in a Vantec HX4 case connect by USB 3.0 It
>> seems a disk I/O error* causes the entire USB connection to reset,
>> causing all the drives to be remapped and screwing up the connections
>> to all the disks in the case.
>>
>> Is this expected behavior for the linux kernel?
>>
>
> I have a drive cage with 4 RAIDed disks in it (two pairs of RAID1)
> connected with USB and I don't see any issues. I am running Jessie and I
> keep my machine up to date (for Jessie, that is :) ).
>
> One thought -- how are you identifying the disks for mounting? If you
> identify them in fstab by UUID, it shouldn't matter if they get remapped
> when the USB infrastructure resets, no?
>
> I see my 4 disks in my drive cage (from the days when they were not
> RAID, admittedly) sometimes come up as Drive 1 = /dev/sdc, Drive 2 =
> /dev/sdd, Drive 3 = /dev/sde, Drive 4 /dev/sdf, but at other times they
> have been Disk 1 = /dev/sde, Disk 2 = /dev/sdf, Disk 3 = /dev/sdc, Disk
> 4 = /dev/sdd, and my machine was none the worse for it.
>
> Mark
>
Reply to: