> Op 9 mei 2016 om 18:36 schreef Sage Weil <sweil@xxxxxxxxxx>:
>
>
> We have a feature (mon_osd_reporter_subtree_level = host) that makes it so
> that if an entire host is down (or whatever the configured hierarchy
> level is) the osds aren't automatically marked out after 5 minutes.
>
> This is confusing on an actual cluster because you see something like
>
> 48/5661 in osds are down
>
> bit it never clears. It's not until you look at the ceph osd tree output
> that you can see why they aren't getting marked out.
>
> It would be great if the health warning said something like
>
> 48/5661 in osds are down
> 1/142 hosts are down (accounting for 48/48 down osds)
>
> and the health detail said something like
>
> host foo is down with 48 OSDs
>
> I think this would be pretty easy to implement given the mon is
> already doing the subtree-based checks.
>
> Thoughts? Any takers?
Seems like a good thing to have. I wouldn't say 'host', since 'mon_osd_reporter_subtree_level' could be set to rack or row if you want to.
Maybe:
480/6720 in osds are down
1/14 of CRUSH type 'rack' are down (accounting for 480/480 down osds)
That seems more logical to me.
Wido
>
> sage
> --
> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
