Re: Subtransaction commits and Hot Standby

Heikki Linnakangas wrote:
> Simon Riggs wrote:
>> Subtransactions cause a couple of problems for Hot Standby:
>
> Do we need to treat subtransactions any differently from normal
> transactions? Just treat all subtransactions as top-level transactions
> until commit, and mark them all as committed when you see the commit
> record for the top-level transaction.

This could lead to inconsistent results -- some of the subtransactions
could be marked as committed while others are still in progress. Unless
we want to be able to atomically mark them all as committed, but I don't
think that's really an option because it could mean holding the clog
lock for a long time, possibly involving I/O of clog pages.

>> Right now we lock and unlock the clog for each committed subtransaction
>> at commit time, which is wasteful. A better scheme:
>> pre-scan the list of xids to derive list of pages
>> if we have just a single page to update
>> {
>> update all entries on page in one action
>> }
>> else
>> {
>> loop thru xids marking them all as subcommitted
>> mark top level transaction committed
>> loop thus xids again marking them all as committed
>> }

> Hmm, I don't see anything immediately wrong with that.

Neither do I.

I wonder if the improved clog API required to mark multiple transactions
as committed at once would be also useful to TransactionIdCommitTree
which is used in regular transaction commit.

--
Alvaro Herrera http://www.CommandPrompt.com/
The PostgreSQL Company - Command Prompt, Inc.