Hi,
Dmitry Anikin schrieb:
>
> Suppose some user issued 'select ... for update', then
> went for coffee-break (to think hard on what he
> really wants to update in that row). Another client
> tries to update the same row and I don't want him to
> wait, just immediately return an error, so he could
> do some other useful task meanwhile. I haven't found
> any no_wait option for locks in the manual :(.
>
> There's a variable innodb_lock_wait_timeout, though, but
> unfortunately I can't assign 0 to it (min. value is 1).
> Still, 1 second time-out can be bearable (although I'd
> appreciate a way to reduce it to zero) but what disturbs
> me is that I've read in the manual that deadlock-removing
What you describe is basically not a deadlock situation!
A deadlock means that two sessions wait for each other in such a way
that neither can proceed before the other one has finished its
"transaction".
Regards,
Frank.
> algorithm aborts transaction which it thinks is most suitable
> for aborting (not last-in-first-aborted). Since time-out
> feature has something to do with deadlocks can I be
> absolutely sure that WAITING transaction will be aborted
> and not that which issued the lock?
> And also it would be fine to have non-destructive means
> to determine whether some row has been locked so I may
> just skip (postpone) some updates without rollback
> of whole transaction. Is it possible?
>
> --
> MySQL General Mailing List
> For list archives: http://lists.mysql.com/mysql
> To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
--
Dr. Frank Ullrich, DBA Netzwerkadministration
Heise Zeitschriften Verlag GmbH & Co KG, Helstorfer Str. 7, D-30625
Hannover
E-Mail: [EMAIL PROTECTED]
Phone: +49 511 5352 587; FAX: +49 511 5352 538
--
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
