Database Reference
In-Depth Information
Sequencing rules : In this section, the level of information that needs to be captured is defined.
The primary parameters in this section are the following:
HANGANALYZE : This defines what level of analysis details should be collected when LTOM
detects a hang on the database server. The various levels are defined in Table 6-4 .
Table 6-4. HANGANALYZE Trace Levels
Level
Description
10
Dump all processes.
5
Dump all processes involved in wait chains.
4
Dump leaf nodes in wait chains.
3
Dump only processes thought to be in a hang.
2
Minimal output.
1
Very minimal output.
SYSTEMSTATE : This defines what level of SYSTEMSTATE dump should be collected. The
default value of 266 (dump all the session history with process stack and current wait
events) for this parameter should be sufficient. Other levels are listed in Table 6-5 .
Table 6-5. SYSTEMSTATE Dump Levels
Level
Description
1
Very basic process information.
10
Most commonly used level and includes object trees for all processes. Includes RAC-related information in
the trace files.
11
Includes dump of any resources not on the freelist. Information is added just after the
BUSY_GLOBAL_CACHE_ELEMENTS ” section of the trace file.
256
Dumps out short stack information for each process.
266
Dump all the session history with process stack and current wait events.
Hang detection directives : This section defines parameters that apply in a general nature to all
scenarios. The important parameters in this section include:
LOOP_HANG_DETECT : TRUE or FALSE . If set to TRUE , LTOM will ignore the default behavior
of detecting only a single hang and then exiting. When set to TRUE , LTOM will not exit
after detecting the first hang but will continue indefinitely to look for other hangs. At each
hang detection LTOM will collect diagnostic traces using the sequencing rules defined
in this file. This parameter should be set with caution, as the automated continuous
collection of diagnostic traces can severely degrade the performance of a system.
RAC_AWARE : It's a Boolean value of TRUE / FALSE and indicates that LTOM understand that
this is a RAC cluster and all required cluster level information should be captured. LTOM
will take all hang analyze and SYSTEMSTATE dumps with the -g flag.
When the hang detection mode is selected/enabled, automatic hang detection proceeds to run in the
background in a silent mode, checking for hangs at predefined intervals set when this option is enabled.
 
 
Search WWH ::




Custom Search