PDT0172 DETECTOR COLLECTION TERMINATING DUE TO HIGH G_COMMON USAGE DB2=DBP4

Products

• Subsystem Analyzer for DB2 for z/OS
• Detector for DB2 for z/OS

Issue/Introduction

Detector  collection was not collecting data and user found the following message in the xmanager log:

PDT0172 DETECTOR COLLECTION TERMINATING DUE TO HIGH G_COMMON USAGE DB2=SSID

Environment

Release : 20.0

Component : CA Detector for DB2 for z/OS

Resolution

The following message is written to the system log and the Xmanager log any time that Detector collection abends and Detector opts not to do an abend restart.  This message will not appear under any other condition:PDT0134 DETECTOR COLLECTION ABEND RESTART NOT ATTEMPTED - INSUFFICIENT TIME - CONSIDER RESTARTING MANUALLY         DB2=DBP4       

Regarding Detector’s abend restart functionality, a second abend restart will occur only if collection is active for one hour after the prior abend restart. This is because the abend restart functionality is not designed to cover up pervasive Detector collection problems. It is designed to restart Detector collection when a rare condition has caused a rare collection abend. When an abend restart is done, several seconds of Detector collection data is missed before collection re-initialization has been completed.

Prior to the abend restart enhancement, rare Detector collection abends had a big impact on those  who run collection for consecutive weeks or months. Some of these collections missed a couple days of Detector collection data before realizing that a rare collection abend had occurred a couple days before. This was the sort of use case we were thinking of when we implemented Detector’s abend restart functionality, now these collections miss several seconds of collection data rather than days of collection data.

In this particular case, the Detector collection terminated 4 minutes after it had been restarted and that was a red flag, telling us that something more than abend restart functionality was needed. This is why the abend restart did not occur.

Something more than Detector abend restart functionality is needed here and PTT(Thread Terminator/Dynamic DSNZPARM) thread monitor is probably the best solution.

The Thread Termination/Dynamic DSNZPARM (PTT) thread monitor provides automated monitoring of threads and cancellation of runaway threads matching specified selection criteria and exceeding specified metric thresholds. PTT would prevent this Detector problem and as well as other problems if used.