Re: Strange Error in Trace File

From: Carl Gohringer <cgohring_at_lucifer>
Date: Wed, 2 Nov 1994 14:40:49 GMT
Message-ID: <CynA42.IEw_at_nl.oracle.com>


psb_at_sambusys.demon.co.uk (Paul Beardsell) writes:
: I'm getting the following error in an Oracle 6.0.36 RDBMS trace file.
: We are running in NOARCHIVELOG mode. What does ``cannot allocate new
: log'' mean? Why does it``need checkpointing''? I have referred to
: the RDBMS Guide but with no success. And if there were an error
: message _number_ I could look it up.
:
: Note that we have three log files and that the error cycles between
: each of the three.
:
: $ tail 3_10070.trc
:
: Advanced to log sequence 670 file /u1/hb/dbs/log1hb.dbf
: Advanced to log sequence 671 file /u1/hb/dbs/log2hb.dbf
: Tue Nov 1 23:50:43 1994
: cannot allocate new log, seq 672, needs checkpointing
: Advanced to log sequence 672 file /u1/hb/dbs/log3hb.dbf
: Advanced to log sequence 673 file /u1/hb/dbs/log1hb.dbf
: Tue Nov 1 23:52:19 1994
: cannot allocate new log, seq 674, needs checkpointing
: Advanced to log sequence 674 file /u1/hb/dbs/log2hb.dbf
: Advanced to log sequence 675 file /u1/hb/dbs/log3hb.dbf
:
At any point in time there is an active redo log, to which redo is currently being written.
When a log fills, the system switches to the next log. When a log switch occurs, a check point automatically occurs, which basically involves writing all dirty blocks in the SGA to disk. While the checkpoint is occuring, both the log you are coming from and the log you are going to are active.(>1 active log, even though only one log is being written to).
In your case, before the checkpoint completes, the second log fills, and another log switch occurs. Cannot switch back to the first log, as it is still active. Basically, until the check point completes, the log swtich will not occur, and your system will hang. When the check point completes, the switch will occur.
Simplest solution is to add another log file. Note, that there is no error condition that is occuring, and nothing in terms of damage that needs fixing.
The only draw back of not fixing this is the occasional hang of the system while you are waiting for the checkpoint to complete.

--
regards,
Carl
+-----------------------------------------------------------------------------+
Carl Gohringer, European Development Centre 
Oracle Park, Bittams Lane, Chertsey, Surrey, England, KT16 9RG
Internet   : cgohring_at_uk.oracle.com    
+-----------------------------------------------------------------------------+
Received on Wed Nov 02 1994 - 15:40:49 CET

Original text of this message