Subject | Re: [firebird-support] Re: what is the most fastest isolation level ? |
---|---|
Author | Ann Harrison |
Post date | 2012-03-03T18:12:59Z |
On Fri, Mar 2, 2012 at 12:48 PM, nathanelrick <nathanelrick@...> wrote:
level and internal. In theory there should be no internal deadlocks, but
we live in practice, not in theory. Internal deadlocks are reported to the
log.
But just to be complete, you do realize that two transactions can't change
the same records at the same time, don't you? It's theoretically unsound.
If you absolutely must do massive updates to the same set of records,
simultaneously, then try to do them in the same order in all transactions.
That reduces the number of deadlocks, causing the second and subsequent
updates to stall on their first change - they'll have to be redone, but at
least they won't have done much before they run into the error.
Good luck,
Ann
[Non-text portions of this message have been removed]
> no one have an idea about what is aDid you check the Firebird lo? There are two types of deadlock - user
>
> deadlock
> Error: 16
>
> ?
>
> the isolation of the transaction was: isc_tpb_read_committed +
> isc_tpb_no_rec_version + wait => normally no deadlock must appear ??
>
level and internal. In theory there should be no internal deadlocks, but
we live in practice, not in theory. Internal deadlocks are reported to the
log.
But just to be complete, you do realize that two transactions can't change
the same records at the same time, don't you? It's theoretically unsound.
If you absolutely must do massive updates to the same set of records,
simultaneously, then try to do them in the same order in all transactions.
That reduces the number of deadlocks, causing the second and subsequent
updates to stall on their first change - they'll have to be redone, but at
least they won't have done much before they run into the error.
Good luck,
Ann
[Non-text portions of this message have been removed]