The thread_stack
section contains analysis instructions around the process stack content :
- The
stack_size_interest
determines the preliminary filtering strategy used to retain the process stacks of interest for the analysis, and discard the ones (the vast majority in many cases) which are considered as useless.
Strategies can be either :known_size_strategy
: the application is well known and the minimum size of stacks of interest is known up front.
This is the recommended approach for any applicative profile.auto_size_stragegy
: the size is determined based on a random set of small recording snapshots (snapshot samples
) which contain a high level of inactivity.
This is the recommended approach for any generic or draft profile.
- The
keep_stacks
section is complementary to the previous one and permits to retain the stacks in staterunning
andlocked
from being discarded. Those are set to true by default. - The
ufo
section determines if the stacks of interest – that did not match one kind of pattern profile – must be isolated for further manual look up.
This permits later on to enrich the pattern profiles.
If activated, the ufo stacks are zipped and placed in the analysis working directory by default.
Set theorder_by
tofrequency
to focus on the ufo stacks appearing the most frequently. - The
ignored_stacks
determines if the discarded stacks must be kept for review.
If activated, the discarded stacks are zipped and placed in the analysis working directory by default. - The sorting key permits to sort the stacks in the analysis – like for example in the Task Sequence sheets – by
thread_name
,thread_id
orrecording
appearance (original behavior).
If not available, the sorting key is defined in the default setup (default_setup.xml) which isthread_id
.
Since Jeyzer 2.6
<thread_stack> <analysis_prefilter> <stack_size_interest> <strategy type="known"> <auto_size_strategy> <scan_coverage snapshot_samples="10" snapshot_sample_percentage="30" percentage="98" stop_if_active_found="false" default_size="1"/> </auto_size_strategy> <known_size_strategy min="2"/> </strategy> <keep_stacks running="true" locked="true"/> </stack_size_interest> </analysis_prefilter> <ufo generation_enabled="true" zip_output_dir="${JEYZER_OUTPUT_DIR}/analysis" order_by="frequency"/> <ignored_stacks generation_enabled="false" zip_output_dir="${JEYZER_OUTPUT_DIR}/analysis"/> <sorting key="thread_name"/> </thread_stack>