jackrabbit-oak-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Chetan Mehrotra <chetan.mehro...@gmail.com>
Subject Re: Conflict handling causes spurious observation events
Date Tue, 23 Jul 2013 09:54:00 GMT
Hi Michael,

On Tue, Jul 23, 2013 at 2:59 PM, Michael Dürig <mduerig@apache.org> wrote:
> It is still possible to have two not branches merging against the same base
> revision. I.e. two branches that are not rebased on top of the current trunk
> and "merge internally". IFAICS it is also necessary for
> NodeStoreBranch.merge() to rebase the branch. I'll give this a try.

I am not sure it would work without going for a cluster level locking.
As the MK does not have any indication that other node is also
committing when it starts the merge it would still merge the changes
internally. So unless MongoMK does not perform a cluster level lock
such a situation cannot be avoided. We can probably make it do that
under some sort of "serialized" transaction mode which can be enabled
as per user requirements.

Chetan Mehrotra

View raw message