Re: TX locks

From: Diego Cutrone <diegocutrone_at_yahoo.com.ar>
Date: Thu, 27 May 2004 13:54:52 -0700
Message-ID: <005b01c4442c$e4617ae0$4d8049c8@DC>

Matthew,

> Why would two transactions need the same TX enqueue? Is it because
> they are attempting to update the same row locally (which I have been=20
> unable to prove or disprove yet)? Is it because they are=20
> both going after the same rows remotely? Is it a lack of available =
> slots
> in the rollback segments (ie, not enough rollback segments)?

What kind of TX enqueue waits are you experiencing? TX-6 or TX-4? TX-6 enqueue waits would indicate that a transaction is trying to lock a row in X mode when that row is currently
being locked by another transaction in an incompatible mode. (most surely another TX-6 lock) . This is an application issue more than a database issue.

TX-4 is a complete different matter. I faced TX-4 problems twice. The first one was related to the lack of free ITL slots in a table (or index), I don't remember. And the second time the TX4 wait was caused when a session was trying to insert a key value (part of an unique index)
in a table after another session had already inserted the same value but had not yet commited. The second session waited in a TX 4 enqueue wait until the first one commited/rollback its transaction.

HTH.
Regards
Diego.

Original Message ----- From: "Adams, Matthew (GE Consumer & Industrial)" <MATT.ADAMS_at_GE.COM> To: <oracle-l_at_freelists.org> Sent: Thursday, May 27, 2004 7:11 AM Subject: TX locks

> I'm having a WHOLE lot of fun trying to track down
> the source of some ORA-2049 (timeout: distributed
> transaction waiting for lock) in a purchased app=20
> called Matrix. I have a number of questions I'm hoping=20
> someone can answer.
>
> Now, according to Metalink, this occurs when a session is=20
> waiting on a TX enqueue that another session is holding AND
> the waiting session is performing a distributed operation
> via a DB link.
>
> Also, according to Metalink (in a different document), TX
> enqueues are taken on particular slots in particular rollback
> segments.
>
>
> If a new connection does, as it's first statment, a read across
> a DB link, is a TX enqueue aquired immediately on a local rollback=20
> segment (as I think it is?)
>
> Why would two transactions need the same TX enqueue? Is it because
> they are attempting to update the same row locally (which I have been=20
> unable to prove or disprove yet)? Is it because they are=20
> both going after the same rows remotely? Is it a lack of available =
> slots
> in the rollback segments (ie, not enough rollback segments)?
>
> None of these scenerios seem very likely in this case, but I'm=20
> grasping at straws here.
>
>
> ----
> Matt Adams - GE Appliances - matt.adams_at_appl.ge.com
> Just once, I wish we would encounter an
> alien menace that wasn't immune to bullets.=20
> - The Brigadier
> ----------------------------------------------------------------
> Please see the official ORACLE-L FAQ: http://www.orafaq.com
> ----------------------------------------------------------------
> To unsubscribe send email to: oracle-l-request_at_freelists.org
> put 'unsubscribe' in the subject line.
> --
> Archives are at http://www.freelists.org/archives/oracle-l/
> FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html
> -----------------------------------------------------------------
>

Please see the official ORACLE-L FAQ: http://www.orafaq.com

To unsubscribe send email to: oracle-l-request_at_freelists.org put 'unsubscribe' in the subject line.

--
Archives are at http://www.freelists.org/archives/oracle-l/
FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------

Received on Thu May 27 2004 - 11:49:30 CDT