Friday, April 4, 2008

11g New Features - ADR

This is a big change in 11g. Now you will not find "alertSID.log" in it's familiar location i.e admin/SID/bdump. It has now got a new location; In fact, Oracle has standardized to store all logs/trace at a single base directory instead of scattering them all over. It's called "Automatic Diagnostic Repository" or "ADR". The parameters "background_dump_dest" and "user_dump_dest" are now deprecated and being replaced by a new parameter called "DIAGNOSTIC_DEST". What more - the alert.log is also available in an xml; Don't worry the text version will still be available.

So from now onwards, you should use "diagnostic_dest" parameter to specify the destination for logs/traces. Oracle creates a sub-directory called "diag" beneath "diagnostic_dest" directory where it will keep all logs/traces. When I say all logs, it is all logs/trace files for all its products/tools including ASM, CRS, Listener etc. This is very great move specifically when CRS stack etc are used; the CRS logs are scattered all over the place.

Here is the new directory structure where you can find all log and trace files :-


Note: I have just expanded and explained key folders in rdbms sub-directory which stores all log and trace files for Oracle database.

sweep -->
stage -->
incpkg -->
incident --> incident dump files
hm --> health monitor files
cdump --> core_dump_dest
metadata --> incident metadata files
lck -->
ir -->
alert --> location for the alert log in xml format
trace --> location for all trace files and text version of alert.log


Some other major improvements are :-
  • Now each critical errors such as ORA-600, ORA-7445 etc. are treated as incidents. Tags the logs with incident id's; Incident alerts are sent to EM and makes entry into alert.log as usual.
  • Incident Packaging service :- IPS enables you to automatically and easily gather the diagnostic data—traces, dumps, health check reports, and more—pertaining to a critical error and package the data into a zip file for transmission to Oracle Support. Awesome indeed.
  • Automatic capture of diagnostic data upon failure
  • Health checks on detecting a critical error
  • V$DIAG_INFO is a new view which lists all ADR locations for the database instance
  • There is a new column called "TRACEFILE" in V$PROCESS which specifies the trace file name for each process.
Above all, 11g comes with a new command line utility called "adrci" to view alert/trace files, package incident trace files so that uploading to oracle support is easy, view health check reports etc. I will explain about this tool in my next post. So all said, the ADR is a great improvement, a great step forward to standardize diagnostic information.

References :- Oracle® Database Administrator's Guide 11g Release 1 (11.1) Part Number B28310-04

1 comment:

Herve Etche said...

Are you sure when you said all logs? Can you double check for the crs logs?