| From: | José Orlando Pereira <jop(at)lsd(dot)di(dot)uminho(dot)pt> |
|---|---|
| To: | Jan Wieck <JanWieck(at)yahoo(dot)com> |
| Cc: | pgsql-hackers(at)postgresql(dot)org, Alfranio Correia <alfranio(at)lsd(dot)di(dot)uminho(dot)pt>, Bruce Momjian <bruce(at)momjian(dot)us>, Theo Schlossnagle <jesus(at)omniti(dot)com>, Jim Nasby <decibel(at)decibel(dot)org> |
| Subject: | Re: Proposal: Commit timestamp |
| Date: | 2007-02-21 16:26:32 |
| Message-ID: | 200702211626.33673.jop@lsd.di.uminho.pt |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Friday 09 February 2007, Jan Wieck wrote:
> I am not sure, I would have to look at what exactly that hook provides.
> The key to a Lamport timestamp is that it is advancing it commit order
> (plus some other things ... of course). If the hook can guarantee that
> the calls are made always in commit order, serialized without any race
> condition possible, it would probably be suitable.
Actually what we do is a bit stronger. We use the commit hook to enforce an
externally defined commit order. In our case, this is defined by a group
communication protocol, which is even allowed to reorder a pair of
transactions originating from the same replica. Therefore, achieving a commit
order that is consistent with a local clock should be straightforward.
Regards,
--
Jose Orlando Pereira
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Bruce Momjian | 2007-02-21 17:06:10 | Re: Enums patch v2 |
| Previous Message | Andrew Dunstan | 2007-02-21 16:22:26 | Re: Column storage positions |