Re: Oracle v7 snapshot too old

micvil_at_lsupoz.apana.org.au (Michael Villanueva) writes:
: We're currently using Oraclr V7 on a Pyramid Nile Machine. My problem is
: whenever I'm doing an update on a large table and one of my peers is doing a
: long query on another table, he's query crashes with the message 'snapshot
: too old'. I'm aware of the data consistency Oracle is trying to provide
: but this query is on another table. Is this an Oracle bug.?

I just did a quick search, and could find no known bugs based on the fairly vague problem description. You might want to give the below a quick read:

   ORA-1555 : SNAPSHOT TOO OLD

  Authors: Chitra Mitra, Ziyad Dahbour, Rama Velpuri

There are various reasons why customers can get the error ORA-1555. Sometimes its due to Rollback segments being too small in size, but there are other reasons. This bulletin is an attempt to give a complete summary of all the situations which would cause an ORA-1555 and how to resolve them. In order to understand the bulletin, one needs to understand some of the internal mechanisms of Oracle, so we start by explaining briefly about read consistency and block cleanouts.

Oracle always enforces statement-level read consistency. This guarantees that the data returned by a single query is consistent with respect to time when the query began. Therefore, a query never sees the data-changes made by transactions that commit during the course of execution of the query.

Oracle uniquely identifies any given point in time by a set of numbers called the System Change Numbers (SCN). So SCN can be defined as the state of the database at any one given point in time. To produce read-consistency, Oracle marks the current SCN as the query enters the execution phase. The query can only see the snapshot of the records as they were at the time of marked SCN.

Oracle uses rollback segments to reconstruct the read-consistent snapshot of the data. Whenever a transaction makes any changes, a snapshot of the record before the changes were made is copied to a rollback segment and the data block header is marked appropriately with the address of the rollback segment block where the changes are recorded. The data block also maintains the SCN of the last committed change to the block.

As the data blocks are read on behalf of the query, only blocks with lower SCN than the query SCN will be read. If a block has uncommitted changes of other transactions or changed data with more recent SCN, then the data is reconstructed using the saved snapshot from the rollback segments. In some rare situations, if RDBMS is not able to reconstruct the snapshot for a long running query, the query results in ORA-1555 error.

A rollback segment maintains the snapshot of the changed data as long as the transaction is still active (commit or rollback has not been issued). Once a transaction is committed, RDBMS marks it with current SCN and the space used by the snapshot becomes available for reuse.

Therefore, ORA-1555 will result if the query is looking for the snapshot which is so old that rollback segment information could not be found becuase of wrap around or overwrite.

SITUATIONS WHERE ORA-1555 ERRORS COMMONLY OCCUR

1. Fewer and smaller rollback segments for a very actively changing database

   If the database has many transactions changing data and commiting very    often, then chance of reusing the space used by a committed transaction    is higher. A long running query then may not be able to reconstruct the    snapshot due to wrap around and overwrite in rollback segments. Larger    rollback segments in this case will reduce the chance of reusing the    committed transaction slots.

2. Corrupted rollback segment

   If the rollback segment is corrupted and could not be read, then a    statement needing to reconstruct a before image snapshot will result    in the error.

3. Fetch across commit

   This is the situation when a query opens a cursor, then loops through    fetching, changing, and committing the records on the same table. In    this scenerio, very often ORA-1555 can result. Let's take the following    example to explain this.

   A cursor was opened at SCN=10. The execution SCN of the query is then    marked as SCN=10. Every fetch by that cursor now need to get the    read-consistent data from SCN=10. The user program is now fetching    x numbers of records, changing them, and committing them. Let's say they    were committed with SCN=20. If a later fetch happens to retrieve a record    which is in one of the previously committed blocks, then the fetch will    see that the SCN there as 20. Since the fetch has to get the snapshot    from SCN=10 it will try to find it in the rollback segments. If it could    rollback sufficiently backwards as previously explained, then it could    reconstruct the snapshot from SCN=10. If not, then it will result in    ORA-1555 error.

   Committing less often which will result in larger rollback segments will    REDUCE the probability of getting 'snapshot too old' error.

4. Fetch across commits with delayed block clean out

   To complicate things, now we see how delayed block clean outs play an    important role in getting this error.

   When a data or index block is modified in the database and the transaction    committed, oracle does a fast commit by marking the transaction as    committed in the rollback segment header but does not clean the datablocks    that were modified. The next transaction which does a select on the modified    blocks will do the actual cleanout of the block. This is known as a    delayed block cleanout.

   Now, take the same scenario as described in previous section. But instead of    assuming one table, lets assume that there are two tables in question.    i.e: the cursor is opened and then in a loop, it fetches from one table and    changes records in another, and commits.    Even though the records are getting committed in another table it could    still cause ORA-1555 because cleanout has not been done on the table    from which the records are being fetched.

   For this case, a full table scan before opening and fetching through the    cursor will help.

   Summary:Fetches across commits as explained in last two cases are not

        supported by ANSI standard.   Accoding to ANSI standard a cursor
        is invalidated when a commit is performed and should be closed
        and reopened. Oracle allows users to do fetch across commits but
        users should be aware that it might result in ORA-1555.

--
regards,
Carl
+-----------------------------------------------------------------------------+
Carl Gohringer, Senior CASE Support Consultant
Oracle UK, The Oracle Centre, The Ring, Bracknell, Berkshire, England, RG121BW
 
Internet   : cgohring_at_uk.oracle.com    
+-----------------------------------------------------------------------------+