What are CA-APAS monitoring capabilities of Adabas Parallel and Cluster Services (ALS) database systems?

Document ID : KB000055683
Last Modified Date : 14/02/2018
Show Technical Document Details

CA-APAS monitors each individual nucleus that makes up the Parallel or Clustered database system as if it were a non-Parallel or Clustered database. It will report all of the activity for that individual nucleus including: number of commands, I/O, response codes, job names, program names, and user id. Everything that can be done in a standard non-Parallel or Clustered environment can be done in a Parallel or Cluster nucleus. Both look the same to CA-APAS and are treated in exactly the same way.

CA-APAS does not provide a display or report of overall activity for all nuclei comprising the Parallel or Clustered database system. If a "Grand Total" or overall report for the Parallel or Clustered database is needed, the user must write the data out for each nucleus to nucleus specific datasets and then use a Natural or other type program to read these datasets and create their own summary reports.

This can be done by starting the same SUMMARIZE request(s) in each nucleus by including it in DBGIN. Each request would contain a standard INTERVAL such as 4, 8, 12, or 24 hours and write the data to a DUAL OUTPUT-FILE. After the interval has occurred, a FLIP command is used to force the interval data to be physically written out to the OUTPUT-FILE. After the DUALCPY process has completed for each nucleus, the user written program is then run to create the overall report for that interval.

The CA-APAS data field - NUCID can be incorporated in the requests as a BY field so the nucleus id number is included in the output data and could be used to sort the data in NUCID sequence if desired.

The major CA-APAS limitation in Parallel and Clustered environments is the inability of Insight to direct Insight commands to a specific NUCID.

Insight is a Natural application. When a user initiates Natural the Adabas SVC decides which NUCID the Natural session is assigned to based on work load algorithms. Users have no control over where the session gets assigned. Once a Natural session is started, there is no way for Insight to direct an Insight command such as SHOW, START, or DISPLAY to a specific NUCID. The commands can only be sent to the NUCID to which the Natural session is assigned.

Some Parallel and Cluster users use the Adabas Set Nucleus command (SN) to logically shutdown all nuclei except the one they want to access. ADACOM will move all the users and work from the nuclei that are logically shutdown to active ones. After the request is started or displayed, the other nuclei are logically started up again.
Here is an example:
(Use the Display command (DIM) to check status)

.+PLI003 SVC=228  DBID=00505 Operator command: DIM                           
.+PLI004  XE44      NUCID UP LO RO .USERS -.CMNDS- LURA=-----               
.+        ADA505N1  50501 Y  Y  G  ------ --------                          
.+        ADA505N2  50502 Y  Y  G  000001 --------                          
.+        ADA505N3  50503 Y  Y  G  000001 --------                          
.+PLI006            * LCL NET DOWN - NO RMT INFO *                          
.+PLI060 SVC=228 DBID=00505 Operator command:SN 50502 DN                    
.+PLI062 Command queued                                                     
.+PLI003 SVC=228  DBID=00505 Operator command:SN 50502 DN                   
.+PLI010                                     Command executed               
.+PLI060 SVC=228 DBID=00505 Operator command:DIM                            
.+PLI062 Command queued                                                     
.+PLI003 SVC=228  DBID=00505 Operator command:DIM                           
.+PLI004  XE44      NUCID UP LO RO .USERS -.CMNDS- LURA=-----               
.+        ADA505N1  50501 Y  Y  G  ------ --------                          
.+        ADA505N1  50501 Y  Y  G  ------ --------                          
.+        ADA505N2  50502 N  Y  G  000001 --------                          
.+        ADA505N3  50503 Y  Y  G  000001 --------                          
.+PLI006            * LCL NET DOWN - NO RMT INFO *                          
.+PLI060 SVC=228 DBID=00505 Operator command:DIM                            
.+PLI003 SVC=228  DBID=00505 Operator command:DIM                           
.+PLI062 Command queued                                                     
.+PLI004  XE44      NUCID UP LO RO .USERS -.CMNDS- LURA=-----               
.+        ADA505N1  50501 Y  Y  G  000001 --------                          
.+        ADA505N2  50502 N  Y  G  ------ --------                          
.+        ADA505N3  50503 Y  Y  G  000001 --------                          

Another potentially confusing aspect is Natural sessions with ADAMODE=2 (the default) are assigned two User Queue Elements (UQEs) and in Parallel and Clustered environments the ADASVC could assign the two UQEs to two different nuclei. The prior example is of an Insight session and the two UQEs were split up between nucleus 2 and nucleus 3.

This may be confusing because in an AOS display the nucleus id for your Natural session could be different from the NUCID CA-APAS displays on the ADARUN screen. It may also differ from the NUCID field in other requests in the same Natural session. This is because one of the UQEs is used for system commands such as AOS commands and the other is used for user commands such as Insight commands.

This division of work can be eliminated by specifying ADAMODE=0 which tells Natural to use only 1 UQE for the session. This, however, could hinder Natural efficiency and through-put and is not recommended. Be aware that Natural sessions could potentially have the two UQEs assigned to different nuclei and the system commands are processed by one nucleus while the user commands are processed by another.