%PSDC-F-UNINSINS error when attempting to restart PSDC

Products

Performance Management for OpenVMS

Issue/Introduction

On one of the nodes in a cluster, we are trying to restart PSDC using the system startup when it gets to the ADVISE COLLECT START it says it didn't start and look at the logfile. The end of the logfile says:

%PSDC-F-UNINSINS, Node #####1 14-AUG-2021 03:12:34.82
Unable to install instrumentation

Here is a full info on that cluster:

<######1>$ advise collect show all

= Performance Schedule File =
===============================================================================
Collection Name // Attributes // Weekly Schedule
===============================================================================
CPD | Monday 0-24
Delete After 7 Days | Tuesday 0-24
HotFile Que. 0.33 | Wednesday 0-24
Interval/sec 120.0 | Thursday 0-24
Start Date 17-OCT-2011 | Friday 0-24
End Date 1-JAN-2999 | Saturday 0-24
Classes: All | Sunday 0-24
Def Path: PSDC$DATABASE:
Def Minimum Free Space: 2500
Def Working Set Size: 2048
-----------------------------------------------------------------------------
-Node---DC Status-------DskFreeSpace--WSquota---Path-------------------------
-----------------------------------------------------------------------------
######0 Stopped 2500 2048 PSDC$DATABASE:
######1 Running 2500 2048 PSDC$DATABASE:
######2 Running 2500 2048 PSDC$DATABASE:
######3 Running 2500 2048 PSDC$DATABASE:
######5 Running 2500 2048 PSDC$DATABASE:
######6 Running 2500 2048 PSDC$DATABASE:
######7 Running 2500 2048 PSDC$DATABASE:
######8 Running 2500 2048 PSDC$DATABASE:
######9 Running 2500 2048 PSDC$DATABASE:
######4 Running 2500 2048 PSDC$DATABASE:

Performance Agent version V3.1-0805 built 6-JUN-2008

We tried shutting down PSDC on all nodes of the cluster, then tried to just startup that one node, but that gave the same results.

Environment

Release : 3.1

Component : PERFORMANCE MANAGEMENT FOR OPENVMS I64

Resolution

When the %PSDC-F-UNINSINS error is found (unable to install instrumentation) the only thing that will allow the Data Collector to work is to reboot the system.

This is due to a design of OpenVMS.

This is the most often culprit that causes the problem.

A PTF is installed and that PTF replaces the executive replaces the executive loadable image named

PSDC$LOADABLE_IMAGE

located in SYS$LOADABLE_IMAGES.

The product was started at least once since the last reboot of the system *and* a new version of the product was installed. This goes hand-in-hand with reason #1 above since we always supply a new version of PSDC$LOADABLE_IMAGE with new versions.

Some other software has placed an address in the memory location referenced via the symbol

EXE$GL_USRUNDWN.

To determine if this is the case, perform the following commands

$ANALYZE/SYSTEM !Get into SDA

SDA>MAP @EXE$GL_USRUNDWN

and the output should look similar to the following if the user rundown handler address is within PSDC$LOADABLE_IMAGE:

Image Base End Image Offset
PSDC$LOADABLE_IMAGE
Short data (read only) FFFFFFFF.88669C00 FFFFFFFF.8866A797 00000000.00070050
SDA>