Smart Problem Classification to Help Oracle Support Resolve Service Requests Faster

AHF diagnostic collection now uses Smart Problem Classification to pinpoint the specific problem for which the diagnostic collection is being performed.

You are often required to collect generic collections for all components over a wide range of times. The logs collected as part of diagnostic collections often reveal evidence of multiple types of problems. Consequently, automated log analysis is limited in its effectiveness because of the significant amount of time required to process and analyze log files.

By intelligently presenting you with a list of detected events relevant to the type of collection being performed, Smart Problem Classification allows you to identify the problem by selecting one from the list.

In addition to recording the type of problem, AHF also records the time and location. This information is made available to Oracle Support to help them resolve Service Requests faster. To ensure that the correct targeted collection is made, you can drill down by problem category if the problem you are looking for is not displayed.

Smart Problem Classification is enabled by default when you run tfactl diagcollect. You can, however, disable it when necessary.

Note:

Currently, Smart Problem Classification is not enabled on systems running AIX and Microsoft Windows operating systems.

How to use Smart Problem Classification

When you initiate a diagnostic collection, Oracle Trace File Analyzer queries events pertinent to the type of collection occurred during the specified time range. If you do not specify the time range, then by default Oracle Trace Files Analyzer queries events occurred duing the last four hours.
Oracle Trace Files Analyzer displays the list of events to pick.
Select one event from the list. Oracle Trace Files Analyzer will initiate a collection to collect data for the selected event.
If you opt to enter a new time range, then Oracle Trace Files Analyzer will prompt you to enter the new time range and will be redirect to step 2.
If you opt to choose to display problem categories, then Oracle Trace Files Analyzer will display a list of categories. Selecting one of them will display sub-categories. After getting the information needed (directed to a problem), Oracle Trace File Analyzer will prompt you to enter the following details:
Time range
Name of the database, if the problem maps to a database.

Setting this will trigger an SRDC for the details provided.

At any point in time, you can exit from the classification process by selecting "X" from the menu.

Note:

The following collection options do not honor Smart Problem Classification:

Collection switches: [-ips, -syslens, -ahf, -awrhtml, -awrtext, -sosreport, -ashhtml, -ashtext]
Collection modules: [-srdc, -insights, -em, -emagenti, -emagent, -oms, -omsi]

Examples Smart Problem Classification

To check if Smart Problem Classification is enabled or disabled:

# tfactl get smartprobclassifier
.---------------------------------.
|              node1              |
+-------------------------+-------+
| Configuration Parameter | Value |
+-------------------------+-------+
| smartprobclassifier     | ON    |
'-------------------------+-------'

To collect diagnostics when Smart Problem Classification is enabled:

# tfactl diagcollect -last 1h

AHF has detected following events from 2022-11-04 12:01:56.768 to 2022-11-04 13:01:56.768
All events are displayed in UTC time zone

Choose an event to perform a diagnostic collection:
1  . 2022-11-04 13:01:43.000 [RDBMS.orcl.orcl1] ORA-00600: internal error code, arguments: [kjb], [ch11], [ch24], [], ...
2  . Show problem categories
3  . Enter a different event time
X  . Exit
Please choose the option [1-3]:2

Problem Categories:
1  . ACFS
2  . ASM Configuration
3  . ASM Errors/Other
4  . ASM Instance Crash
5  . CRS Client
6  . CRS Errors/Other
7  . Clusterware Installation
8  . Clusterware Patching
9  . Clusterware Startup
10 . Clusterware Upgrade
11 . Database Corruption
12 . Database Errors/Other
13 . Database Install
14 . Database Instance Eviction/Crash
15 . Database Internal Error
16 . Database Memory
17 . Database Patching
18 . Database Performance
19 . Database RMAN
20 . Database Storage (ASM)
21 . Database Streams/AQ
22 . Database Upgrade
23 . Dataguard
24 . GoldenGate
25 . Node Eviction/Reboot
26 . Problem not listed, provide problem description
X  . Exit
Please select the category of your problem [1-26]:
...
...

To bypass Smart Problem Classification:

# tfactl diagcollect -last 1h -noclassify
Collecting data for all nodes
TFA is using system timezone for collection, All times shown in UTC.
Collection Id: 20221104125517stbm000004-vm15

To bypass Smart Problem Classification, use the flags -silent and -noclassify.

# tfactl diagcollect -last 1h -silent
Smart Problem Classifier is ON. Since -silent is passed, Problem Classifier is not processing the request.

To disable Smart Problem Classification:

# tfactl set smartprobclassifier=off
Successfully set smartprobclassifier=OFF
.---------------------------------.
|               node1             |
+-------------------------+-------+
| Configuration Parameter | Value |
+-------------------------+-------+
| smartprobclassifier     | OFF   |
'-------------------------+-------'

To enable Smart Problem Classification:

# tfactl set smartprobclassifier=on
Successfully set smartprobclassifier=ON
.---------------------------------.
|               node1             |
+-------------------------+-------+
| Configuration Parameter | Value |
+-------------------------+-------+
| smartprobclassifier     | ON    |
'-------------------------+-------'