DX Application Performance Management

FYI I found that even with Postgres on same server as EM the SP1 HF install in EM_HOME did not find it so I just re-ran the HF install from /opt/PostgreSQL-9.6.2/:

===

...

PostgreSQL not found - don't forget to execute this patch for PostgreSQL too!
Hotfix 10.7.0-HF08 was applied successfully
# cp APM10.7.0.90SP1.jar /opt/PostgreSQL-9.6.2/
# cd /opt/PostgreSQL-9.6.2/
# /root/Introscope10.7.0.45/jre/bin/java -jar APM10.7.0.90SP1.jar
Initializing hotfix
Creating hotfix folder hotfix/10.7.0-HF08
creating directory structure hotfix/10.7.0-HF08
Applying hotfix scenario
EM not found - don't forget to execute this patch in EM directory too!
Webview not found - don't forget to execute this patch in Webview directory too!
Workstation not found - don't forget to execute this patch in Workstation directory too!
APMSqlServer not found - don't forget to execute this patch in APMSqlServer directory too!
CommandCenter not found - don't forget to execute this patch in CommandCenter directory too!
PostgreSQL found - proceeding with patching
creating directory hotfix/10.7.0-HF08/Database
moving /opt/PostgreSQL-9.6.2/pgAdmin 4
moving file /opt/PostgreSQL-9.6.2/include/libxslt/xsltutils.h to hotfix/10.7.0-HF08/Database/xsltutils.h
moving file /opt/PostgreSQL-9.6.2/include/libxml2/libxml/xmlunicode.h to hotfix/10.7.0-HF08/Database/xmlunicode.h
Hotfix 10.7.0-HF08 was applied successfully

===

There is a new property introduced in APM 10.7 SP1 causing some MM not to load

A KB is in process:

https://ca--c.na60.visual.force.com/apex/CustomKnowledgeEdit?articleId=ka00c000000yfSYAAY

Description

After applying SP1 to 10.7 cluster some Management Modules are not loading,

Example from log:

6/11/18 09:50:42.167 AM EDT [INFO] [PO Async Executor] [Manager.ManagementModule] Loading Management Module "Wily-CA"
6/11/18 09:50:42.171 AM EDT [ERROR] [PO Async Executor] [Manager.ManagementModule] Unable to load the Alert named "Wily Engine Memory" in Management Module "Wily-CA" - 'The Alert "Wily Engine Memory" cannot have danger periods of at least 40 out of 40 or caution periods of at least 40 out of 40.'.
6/11/18 09:50:42.171 AM EDT [ERROR] [PO Async Executor] [Manager.ManagementModule] Unable to load the Alert named "Wily Machine CPU Used - Percent" in Management Module "Wily -CA" - 'The Alert "Wily Machine CPU Used - Percent" cannot have danger periods of at least 60 out of 60 or caution periods of at least 20 out of 20.'.

Cause:

introduced a new property called "introscope.enterprisemanager.alerts.maxPeriods". Any Alert Danger/Caution period larger than maxPeriods will be considered invalid. The default value is 20. 40>20, so it fails. And it's a chained reaction: one Alert fail to load causing other Elements referencing this Alert also fail to load, and so on and so forth. For customer seems a lot of MM/Alert has been impacted.

Resolution:

Set maxPeriods to be a big enough number to cover all existing Alerts. Otherwise customer would have to change definition for all their impacted Alerts to reduce the period number, which is not ideal...

1. Why this new property was added?
It’s a new clamp to avoid defining Alert with very high periods that could potential increase memory overhead.

2. What is the reason of the value 20?
I think value 20 was just an initial thought of default value. We’ll probably increase it in later releases.

3. Is there a way to know how many MM has the Alerts danger/caution periods numbers?
Please engage with CA Professional Services who may be able to create a script for you.

4. How much the value can be raised? Is there any limit?
There is no hard limit.

5. Will it impact any EM performance?
The clamp itself won’t. But if the clamp is set very high, and huge number of Alerts were defined with very high periods, then memory overhead might be high and impact EM performance.

6. Is a hot property? Or EM restart is required.
The property itself is hot, but changing the clamp won’t trigger reloading any Alerts that failed to load at startup time.

Hi Richard,

SP1 fixes up some Postgres vulnerabilities

The  "HOME directory" of the "PostgreSQL database" would be your install directory for the 10.7 version of Postgres 9.2.6 which defaults to /opt/PostgreSQL-9.6.2.

Further up the README it states:

===

If the PostgreSQL database is not installed along with the EM, then the customer
needs to execute the patch from the PostgreSQL Install directory. The pgAdmin 4 folder
is deleted from the PostgreSQL installation or upgrade due to security vulnerabilities
and is therefore no longer accessible as part of PostgreSQL installation. Users who
require pgAdmin to view the PostgreSQL data can download supported pgAdmin
from the pgadmin website: https://www.pgadmin.org/download/

===

So running the Hot Fix APM10.7.0.90SP1.jar from PostgreSQL database HOME directory is only needed if your Postgres APM DB is NOT installed on the same server as EM/MOM which is what is normally recommended for a Production Environment.

If Postgres is installed on same server as EM/MOM then running the the Hot Fix APM10.7.0.90SP1.jar from EM HOME directory will also update Postgres.

Hope that helps.

Regards,

Lynn

The instructions provided in the SP1 README state: "Copy APM10.7.0.90SP1.jar to the HOME directory of the product components". Furthermore, "PostgreSQL database" is included in the list components to be updated.

Question: Which directory is considered the "HOME directory" of the "PostgreSQL database" component?

More information on new SPM10.7  SP1 parameters introduced please see: 10.7 SP1 Configuration Changes & Important Post Install Notes 

Not in the Readme is that APM 10.7.0 SP 1 supports authentication for MQ Monitor Agent . See https://comm.support.ca.com/kb/mq-monitor-agent-not-passing-credentials-with-mqcsp-as-requested-by-our-mq-admins/KB000097920  for details.