(Very) good question.
It's hard for me to judge at this point because my application isn't
customer-facing yet and I don't have any metrics to measure against. So
by that yard stick, it's certainly good enough for now. I just antsy
about deadlocks and scalability problems once I go public. FYI:
Converting the database and creation script to Postgresql took around 4
days. With the migration nearly complete I'm not sure how to proceed
(whether to go back to H2 or continue with Postgresql).
I'll try to invest more energy tracking down this AUTO_INCREMENT problem
but thus far I've only been able to reproduce it 1 out of every 20 runs.
I've produced a testcase that inserted all possible values in the
SMALLINT range and the problem did not occur so at this point it's
looking increasingly likely that this is a race condition that only
occurs when two threads try to insert at the same time. I'll keep on trying.
Gili
On 17/09/2014 2:17 AM, Noel Grandin wrote:
Why do you need to use READ_COMMITTED at all? Are we not fast enough
without it?
On 2014-09-16 10:57 PM, cowwoc wrote:
Hi Noel,
This is already I what I do (in SERIALIZABLE mode). If you were to
run the same in READ_COMMITTED mode you could get a
conflict on that last INSERT. The problem with DUPLICATE_KEY_1 is the
following:
--
You received this message because you are subscribed to the Google Groups "H2
Database" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/h2-database.
For more options, visit https://groups.google.com/d/optout.
