Re: ld vs. dkctl in netbsd-8

[David Brownlee <abs%absd.org@localhost>]

On Thu, 29 Nov 2018 at 09:00, Louis Guillaume <louis%zabrico.com@localhost> wrote:
>
> I just noticed this message in the console after a reboot:
>
>    raid2: get cache for component /dev/dk1 failed
>
> which is a bit troubling, considering I didn't get a similar message for
> dk0, which is the other half of that raid1 pair (on raidframe).
>
> These drives are on separate (slightly different) twa controllers, using
> the ld driver. I recently updated this driver due to MP_SAFE crashes by
> applying the patch in this PR (the latest patch)...
>
>    http://gnats.netbsd.org/cgi-bin/query-pr-single.pl?number=53628
>
> Here's an excerpt from the dmesg. Is this cache discrepancy a potential
> performance hit or should I be otherwise concerned?
>
> Any thoughts or help would be great.

As an empirical check I'd be inclined to perform some write tests to a
filesystem on the RAID and see if the disks report similar time
utilisation from 'iostat -x 5'.
Probably just dd'ing /dev/zero to a file would be fine, though you
could also run a disk test like bonnie++.

David