Re: Interpreting USED_UREC from V$TRANSACTION
Date: Fri, 6 Jul 2012 10:42:02 -0700 (PDT)
Message-ID: <1341596522.7866.YahooMailNeo_at_web121606.mail.ne1.yahoo.com>
One possibility is the value generated is too large for a signed integer variable causing the value to go negative -- it was a problem a while back in one of the Solaris kernels (and Linux, I believe) where an uptime of greater than 260+ days caused counters to go negative. David Fitzjarrell
From: Herring Dave - dherri <Dave.Herring_at_acxiom.com> To: "Oracle-L (oracle-l_at_freelists.org)" <oracle-l_at_freelists.org> Sent: Friday, July 6, 2012 11:20 AM
Subject: Interpreting USED_UREC from V$TRANSACTION
Folks,
I'm monitoring a VERY large rollback operation and could use some help in understanding conflicting values for USED_UREC in V$TRANSACTION.
This is for 11.2.0.3 on RHEL 5.6. The transaction was a DROP COLUMN against a 1.3 TB table which I believe ran out of undo after 50+ hrs. The SQL was generated from a data modeling tool and even after warning about runtimes they wanted the command run exactly as generated.
Now that the DROP COLUMN command has been rolling back for 5 days, I'm getting pretty confused on what V$TRANSACTION is telling me, specifically USED_UREC. Periodically over the last few days I've been querying this view and the values decreased way below zero and are now back about 800 million. In all cases each check a little later showed the values were decreasing so I knew it was still rolling back:
Snapshot Date USED_UREC
02-JUL-12 10:32:12 390,963,088 03-JUL-12 18:41:54 -903,130,274 06-JUL-12 12:57:17 784,607,392
As the value for USED_UREC decreased near 0 I figured that'd be it, but then it seemed to go indefinitely below 0, then some point in the past 24 hrs. the value was reset about 800 million. Like I said, repeated checks so the value to be decreasing and also a DECODE(BITAND(t.flag,128), 0, NULL, 'Rollback') returns "Rollback".
So how can the value go negative and then jump back to 800 million? I'm pretty sure it didn't "wrap" and transactions can't span undo segments. This also makes it VERY difficult to figure out how long this will take.
The best I've come up with is comparing "redo entries" stat from AWR before the rollback with "rollback changes - undo records" stat in AWR during this rollback process. Should they eventually be equal, assuming little to no other activity on the DB?
DAVID HERRING
DBA
Acxiom Corporation
EML dave.herring_at_acxiom.com<mailto:dave.herring_at_acxiom.com> TEL 630.944.4762 MBL 630.430.5988
1501 Opus Pl, Downers Grove, IL 60515, USA WWW.ACXIOM.COM<http://www.acxiom.com/>
[cid:image001.png_at_01CD5B6F.7DC25E40]<http://www.facebook.com/acxiomcor> [cid:image002.png_at_01CD5B6F.7DC25E40] <http://www.linkedin.com/groupRegistration?gid)0173> [cid:image003.png_at_01CD5B6F.7DC25E40] <http://twitter.com/acxio>
[Description: cid:image004.png_at_01CB84F1.26214350]
The information contained in this communication is confidential, is intended only for the use of the recipient named above, and may be legally privileged.
If the reader of this message is not the intended recipient, you are hereby notified that any dissemination, distribution or copying of this communication is strictly prohibited.
If you have received this communication in error, please resend this communication to the sender and delete the original message or any copy of it from your computer system.
Thank You.
-- http://www.freelists.org/webpage/oracle-l -- http://www.freelists.org/webpage/oracle-lReceived on Fri Jul 06 2012 - 12:42:02 CDT