On Tue, Jan 25, 2022 at 7:52 AM Mariusz Tkaczyk
<mariusz.tkaczyk@xxxxxxxxxxxxxxx> wrote:
>
> On Fri, 17 Dec 2021 09:02:38 +0100
> Mariusz Tkaczyk <mariusz.tkaczyk@xxxxxxxxxxxxxxx> wrote:
>
> > Hi Song,
> >
> > On Thu, 16 Dec 2021 16:52:23 -0800
> > Song Liu <song@xxxxxxxxxx> wrote:
> > > > Mariusz Tkaczyk (3):
> > > > raid0, linear, md: add error_handlers for raid0 and linear
> > > > md: Set MD_BROKEN for RAID1 and RAID10
> > > > raid5: introduce MD_BROKEN
> > >
> > > The set looks good to me. The only concern is that we changed some
> > > messages. While dmesg is not a stable API, I believe there are
> > > people grep on it to detect errors.
> > > Therefore, please try to keep these messages same as before (as much
> > > as possible).
> >
> > Will do.
> > After sending it, I realized that my approach is not correct when
> > mddev->fail_last_dev is on. MD_BRKOEN should be set even if we agree
> > to remove the "last" drive. I will fix it too.
> >
>
> Hi Song,
> For raid0 and linear i added new messages so it shouldn't be a problem.
> I added one message in raid5 for failed state:
> + pr_crit("md/raid:%s: Cannot continue on %d devices.\n",
>
> Do you want to remove it?
> Other errors are same. Order is also preserved.
I think we should print in this case. How about we change it to something
similar to the one in raid5_run():
pr_crit("md/raid:%s: Cannot continue operation (%d/%d
failed)\n",
mdname(mddev), mddev->degraded, conf->raid_disks);
Thanks,
Song
