Bug 216809 - nvme nvme0: I/O 0 (I/O Cmd) QID 1 timeout, aborting

[Keith Busch]

On Fri, Oct 13, 2023 at 03:25:37AM +0200, Andreas Thalhammer wrote:
> I'd like to make you aware of
> Bug 216809 - nvme nvme0: I/O 0 (I/O Cmd) QID 1 timeout, aborting
> 
> https://bugzilla.kernel.org/show_bug.cgi?id=216809
> 
> I personally don't see the reported errors AND system freeze on my own
> systems, but the reporters do, which are:
> jfhart085 at gmail.com
> Nelson G <konoha02 at yahoo.com>

Does the device just stop responding entirely, or is it just reponding
exceptionally slowly? Output from something like "iostat -X 1" that
includes the period timeouts occur would be useful to see what workload
is being thrown at the device, and how it is handling it.

If the device is responding very slowly, suggestions I might have
include:

  a. Enable discards if you've disabled them
  b. Disable discards if you've enabled them
  c. If not already enabled, use an io scheduler, like mq-deadline or kyber

If the device isn't responding at all, which appears to be the case in
one of the dmesg's from the bz, then suggestion is report a bug to the
device vendor.