Oracle-L: RE: Rollback and what table's DML caused an ORA-1555?

From: <Jared.Still_at_radisys.com>
Date: Mon, 07 Apr 2003 11:43:42 -0800
Message-ID: <F001.0057C20F.20030407114342@fatcity.com>

Rich,

Reset optimal like this:

alter rollback segment <SEGMENT_NAME> storage ( optimal NULL )'

I sometime run a script to set optimal on all rbs segments, shrink the segments,
then reset optimal to null.

This has proven to be useful when it's known ahead of time that some large rbs eating operation is about to be take place.

Be forewarned: it can cause ORA-1555 on current queries.

Jared

"Jesse, Rich" <Rich.Jesse_at_qtiworld.com>
Sent by: root_at_fatcity.com
04/04/2003 01:48 PM
Please respond to ORACLE-L

        To:     Multiple recipients of list ORACLE-L <ORACLE-L_at_fatcity.com>
        cc: 
        Subject:        RE: Rollback and what table's DML caused an ORA-1555?

Excellent advice, Gaja! Yes, we do have OPTIMAL set. From the docs I don't
see a way to remove the OPTIMAL setting once used. I assume this means new
RBSs?

Rich

Rich Jesse                        System/Database Administrator
rich.jesse_at_qtiworld.com           Quad/Tech International, Sussex, WI USA

-----Original Message-----
Sent: Friday, April 04, 2003 12:54 PM
To: Multiple recipients of list ORACLE-L

Rich,

The overcommiting is a definite suspect and is worth looking into. ALso, do you by chance have OPTIMAL set on your rollback segments? If so, I'd suggest you remove the OPTIMAL clause and try again. In my experience, I have had my share of hassles with OPTIMAL. Even when it was sized 'reasonably large' (way above INITIAL * MINEXTENTS) for the application.

OPTIMAL does increase the probability of ORA-1555, as extents of your rollback segments that have the "before images" of your transactions, can get dropped, while your queries are left "high and dry". Just something to check.

The downside of not setting OPTIMAL is more disk usage for your rollback segments. In the bigger scheme of things, it may be much cheaper to eat the disk cost (which is probably a few dollars) than to deal with queries failing with ORA-1555s. If you are so strapped for space, then you can "hand shrink" the rollback segments using the ALTER rollback segment xxx SHRINK command via a job, at a time when no queries are running.

If you don't have OPTIMAL set, and if you are using any READ-ONLY tablespaces that were put in READ-ONLY mode recently, then try the following:

Set those tablespaces back to READ-WRITE mode
Write a sql-generating-sql script that performs a select count(*) on all the objects in the tablespace
Set the tablespaces back to READ-ONLY.

The above exercise will force a block-cleanout (if required) on all the objects, which may not have occured before the tablespace was originally put in READ-ONLY mode.

Hope this helps,

Gaja

"Jesse, Rich" <Rich.Jesse_at_qtiworld.com> wrote:
> Hey all,
>
> Fighting with a lot of ORA-1555s lately on 8.1.7.4
> on HP/UX. Most of them
> are now coming from long-running Business Objects
> (B.O.) queries against our
> OLTP DB. I think I need to recreate the RBS
> tablespace (currently 1MB
> extents in LMT), but until I can get time to do
> that, I'd like to approach
> this from the application side, where I think the
> majority of the problem is
> occurring. I've been tracking TPM based on "user
> commits" in V$SYSSTAT and
> we spiked at 20K TPM just before the B.O. query in
> question ORA-1555'd.
> From STATSPACK reports, I think the most likely
> cause for this is a COMMIT
> for every DML in a batch job. From what I've read,
> including MetaLink
> 40689.1, this over committing is one potential cause
> of ORA-1555s.
>
> In order to narrow down the problem, I've turned on
> event 1555 in the
> instance. Is it possible to determine what
> table(s)' DML is causing the
> ORA-1555 based on the trace file? I have the last
> wait state, which happens
> to be "db file sequential read", but I don't know if
> there's any
> correlation. If there is, I should be able to
> determine which table by the
> file# and block# given in the trace. Is this
> correct?
>
> Also, if the over-committing process is not doing
> any DML on the tables of
> the B.O. query, is it still a possible suspect of
> causing the ORA-1555
> because of the potential of overwriting another
> process' RBS?
>
> Damn. I was hoping to be at 9i before I had to deal
> with RBSs... :)
>
>
> Rich
>
> Rich Jesse System/Database
> Administrator
> rich.jesse_at_qtiworld.com Quad/Tech
> International, Sussex, WI USA
-- Please see the official ORACLE-L FAQ: http://www.orafaq.net -- Author: Jesse, Rich INET: Rich.Jesse_at_qtiworld.com

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------

To REMOVE yourself from this mailing list, send an E-Mail message to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing).

-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.net
-- 
Author: 
  INET: Jared.Still_at_radisys.com

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).

Received on Mon Apr 07 2003 - 14:43:42 CDT