On Fri, Nov 27 2015, Guoqing Jiang wrote: > Hi Neil, > > On 11/27/2015 01:42 PM, NeilBrown wrote: >> On Fri, Nov 20 2015, Guoqing Jiang <gqjiang@xxxxxxxx> wrote: >> >>> To change the bitmap from clustered to none, we also >>> need to clean related info about cluster from sb, such >>> as version, nodes and cluster_name. >> I've applied the other two (though I removed the __func__ and __LINE__ >> From error messages - they are just noise: the source of the message can >> easily be found with out them - I should remove all the others). >> >> However I haven't applied this because there doesn't seem to be any >> point. >> Once your told that main md superblock that there is no bitmap, it >> doesn't really matter what data is in the superblock of that >> non-existent bitmap - it will never be looked at. Will it? > mdadm could still get nodes and cluster_name after change > bitmap to none, so some cmds like "mdadm -X /dev/sda" could > still see clustered info, it seems a little confused for user. In that case we should fix mdadm to not report a bitmap if MD_FEATURE_BITMAP_OFFSET isn't set (though maybe still report it if --force is given). i.e. if this is an issue, it is not specific to clustered-bitmaps. NeilBrown
Attachment:
signature.asc
Description: PGP signature
