Posted on

Where to find Oracle clusterware log files.

To make administration and problem diagnosis easier in Oracle Clusterware, there is a single consolidated directory structure for the main clusterware log files.

 

This is found under $GRID_HOME/log/[hostname]/

 

The CRSD logfile – crsd.log is stored in the /crsd/ directory and is archived every 10mb.

 

The CSS logfiles – cssd.log is stored in the /cssd/ directory. This is archived every 20mb.

 

EVM logs are under /evmd/

 

SRVM (srvctl) is found along with the OCR logfiles under /client/ and also under $ORACLE_HOME/log/[hostname]/client/

 

Posted on

Oracle Architecture Part 2

[embedplusvideo height=”328″ width=”584″ standard=”http://www.youtube.com/v/wmc9uxNW99E?fs=1″ vars=”ytid=wmc9uxNW99E&width=584&height=328&start=&stop=&rs=w&hd=0&autoplay=0&react=1&chapters=&notes=” id=”ep5503″ /]

Check out part 1 first.

Notes:

Online Redo Logs

  • Contains Redo and Undo info
  • DML
  • Inserts/Deletes
  • DDL – Create alter drop
  • Commits
  • Need at least 2 files/groups
  • Should be multiplexed each on different devices and controlers
  • When each file fils, next one is started and old one is archived (if in Archivelogmode)

An Oracle database is comprised of datafiles, control files
and redo logs.

The second part is the memory – the SGA (system global area
or shared global memory)

Database Buffer Cache

Comprised of Oracle data blocks

There are 4 main types of data blocks, which are the same as the 4 types of data mentioned earlier:

  • Data
  • Index
  • Rollback
  • Temporary

Part 3…

Posted on

Oracle Log Switches and Checkpoints

When a redo log file fills up it switches to the next file. This is called a log switch.

The log switch always causes a checkpoint to occur. This flushes all dirty buffers from the log cache. (writes modified data to disk)

If archiving is on, the old log file id archived.

The new log file is given a log sequence number which is also given to the archived log file.

 

Controlint the checkpoint

LOG_CHECKPOINT_INTERVAL

set to the number of operating system blocks that can be used before a checkpoint occurs.

LOG_CHECKPOINTS_TO_ALERT

Writes a message to the alert log when a checkpoint occurs.

This can take up a lot of space!

LOG_CHECKPOINT_TIMEOUT

Set the time in seconds bertween checkpoints

 

Force a checkpoint

ALTER SYSTEM CHECKPOINT

You may do this if there is a risk of the system failing.

 

Force a log switch

ALTER SYSTEM SWITCH LOGFILE;

you will rarely need to do this.