HI Ethan,

This looks really good. I do have a question though.

On line 876 it's setting be_created if we get a failure back from 
be_clone_fs_callback. If this fails then we haven't created the BE
but we're still setting be_created, doing the goto done and
attempting to clean-up the BE. I can understand if we've at least
created one of a set of needed snapshots (or clones) we need to clean
up but what happens in the case where no snapshots (or clones) were
created? I guess that gets covered in _be_destroy in that the first
zfs_open will fail?

Anyway the changes look fine, I just wanted to make sure this scenario
was covered.

Thanks,
-evan

Ethan Quach wrote:
> Two more fixes in this webrev.  The first defect fixes the case where
> creating a BE will silently ignore failures in creating corresponding
> zones in the new BE.  It will now stop the creation process and report
> error.  be_copy() will now also cleanup a partially created boot
> environment upon any failure that happens during the creation
> process instead of leaving it lying around.
> 
> The second defect is a regression due to 1197.  The additional test
> case that caught this was doing a pkg image-update from a system
> where the pool is running less then the required pool version that
> supports snapshot user properties.
> 
> 
> 
> Webrev:
> ----------
> http://cr.opensolaris.org/~equach/webrev.3844.3889
> 
> Defects:
> ----------
> 3884 be_copy_zones() silently ignores zones
> http://defect.opensolaris.org/bz/show_bug.cgi?id=3844
> 
> 3889 be_create_snapshot fails when running pkg image-update
> http://defect.opensolaris.org/bz/show_bug.cgi?id=3889
> 
> 
> thanks,
> -ethan
> 
> _______________________________________________
> caiman-discuss mailing list
> caiman-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss


Reply via email to