Re: Seemingly random nvme (nda) write error on new drive (retries exhausted)

On 6/8/23 05:48, Warner Losh wrote:
>
>
> On Thu, Jun 8, 2023, 4:35 AM Rebecca Cran <rebecca@bsdio.com> wrote:
>
>     It's ZFS, using the default options when creating it via the FreeBSD
>     installer so I presume TRIM is enabled. Without a reliable way to
>     reproduce the error I'm not sure disabling TRIM will help at the
>     moment.
>
>     I don't think there's any newer firmware for it.
>
>
> pci gen 4 has a highter error rate so that needs to be managed with 
> retries.  There's a whole protocol to do that which linux implements. 
> I suspect the time has come for us to do so too. There's some code 
> floating around I'll have to track down.

Thanks. I dropped the configuration down to PCIe gen 3 and the errors 
have so far gone away.


nda0: nvme version 1.3 x8 (max x8) lanes PCIe Gen3 (max Gen4) link
nda1: nvme version 1.3 x4 (max x4) lanes PCIe Gen3 (max Gen4) link

-- 
Rebecca Cran