Re: SCSi issues after 10.0 update

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

On Fri, 31 May 2024 at 00:03, Riccardo Mottola
<riccardo.mottola%libero.it@localhost> wrote:
>
> Hi guys,
>
> David Brownlee wrote:
> > (Adding a little more detail in case it helps:)
> >
> > One approach would be:
> > 1) find a test which always fails on netbsd-10 (maybe
> > pkgsrc/benchmarks/bonnie++ or similar)
> > 2) boot a netbsd-9 install and confirm ir runs without issue
> > 3) cross build of kernel from somewhere between
> > 4) test boot the new kernel with the netbsd-9 install, and loop back
> > to 3) either closer to netbsd-9 or netbsd-10
> >
> > I'm happy to help with 3) if you don't have a particularly fast build
> > box - I can run up a set of test kernels (say one per month between -9
> > and -10) and you can binary search across them
>
> thanks for the proposals... however I fail at 1) !
> I tried doing again the same thing: sharing over NFS with other boxen
> building and compiling heavy stuff on the same system - it stayed up for
> several days.
>
> Then I thought of stress-testing. I launched bonnie++ several times on
> on edisk, on the other and then launched two instances in parallel on
> both disks, repeatedly.
> No problems!
>
> I really wonder what happened the first time... what error it was? For
> sure the system acceptable stable. Also, this way, I cannot say 9.3 was
> unaffected.. albeit in A year or so I never had to reboot forcibly the
> system it was very stable.
>
> I suppose that the disks aren't in a too bad shape and all seems stable
> even with torturing, bonnie writes gigabytes and thousands of files.
>
> The only thing that bonnie doesn't dest is concurrency with other
> devices - eg. high network activity with concurrent high disk activity.
>
> I could try running bonnie++ over NFS??

If you saw the initial issues with over NFS then bonnie++ over NFS
could be an excellent test

(Many moons ago I used bonnie running on a handful of render clients
to show a proposed migration to Origin 2000 servers with Fore Systems
ATM drivers was going to be a disaster - repeatably triggering a panic
within a minute or so)

David