[Gluster-Maintainers] [Gluster-devel] Release 3.10.1: Scheduled for the 30th of March
srangana at redhat.com
Mon Mar 27 21:30:28 UTC 2017
On 03/27/2017 03:25 PM, Gandalf Corvotempesta wrote:
> 2017-03-27 18:59 GMT+02:00 Shyam <srangana at redhat.com>:
>> 1) Are there any pending *blocker* bugs that need to be tracked for 3.10.1?
>> If so mark them against the provided tracker  as blockers for the
>> release, or at the very least post them as a response to this mail
> I think that file corruption when sharding is enabled during a rebalance *must*
> be considered blocker for any release.
I agree the problem is critical and needs a resolution soon.
But, till we have a resolution for the same, we cannot hold other
bugs/releases at bay. In this case, it forms a good limitation/known
issue, for the release, and can be added to the release-notes, so that
there is no confusion created anywhere.
Further, we could possibly disable rebalance when shard is in the mix,
so that new users (or those that miss the release notes) do not shoot
them selves in the foot.
> If you still continue to release new version with the same bugs around,
> you only create confusion to users, willing to think that bugs
> (especially the critical ones)
> are fixed.
Our intention is never to confuse anyone about a release, in stating
that a fix is available when it is not. The release notes should be
clear in that regard.
For *this* bug I think Krutika provided a detailed enough response on
why we thought it was fixed, and where we stand now, so I am not going
to restate that again, as her response is detailed well.
> I've never ever seen a software releasing new versions ignoring existing bugs.
Are you stating we should fix *all* *pre-existing* critical(?) bugs
before releasing? If so the short answer is, that will not happen.
(regressions introduced and discovered within a release cycle are a
different matter though, and will go through enough scrutiny before a
decision to release or postpone the release is taken)
To repeat myself, I do understand that this is a critical bug, and we
apologize for the problem at hand, but as stated in the other mail
thread, this is proving difficult to reproduce and root cause. Any help
towards the same would be much appreciated, than delaying releases for
other fixes that are already in the code.
More information about the maintainers