Author: Artemk
Posted: Tue Nov 22, 2016 8:59 pm (GMT 5.5)
Forgot to tell about another observation.
(Some clarification about what does the program do:
BMP takes the data from prepared GSAM input file. one record consists of 3 fields: DSN=44char, Type=3-char, data=1917char and each record is inserted in IMS database like 3 segment - root, child, child of child.
I described it so detailed to show that each input record has the same structure, and causes the insertion of the same segmnets of same size.)
And back to BMP. I start the BMP program with NBA=2000. The first FW I get after 2000 input records - as expected. Make checkpoint. Next FW appears after 19xx , the next after even smaller number, and so on.
For example, 100th checkpoint happens 1800 records read after 99th. But on my input data, BMP program ends successfully.
If I use NBA=8000, the behaviour is the same. Sooner or later the interval between checkpoints decreases until about 4000, and 878 abend happens.
This abend says that GetMain failed, so I decided to check what happens with CSA and ECSA. Both of them have a plenty of place - IMS have 1GB, uses 100-160MB, some times up to 600MB. ECSA has about 4500MB, and 2GB used. So, seems that no real lack of memory.
Posted: Tue Nov 22, 2016 8:59 pm (GMT 5.5)
Forgot to tell about another observation.
(Some clarification about what does the program do:
BMP takes the data from prepared GSAM input file. one record consists of 3 fields: DSN=44char, Type=3-char, data=1917char and each record is inserted in IMS database like 3 segment - root, child, child of child.
I described it so detailed to show that each input record has the same structure, and causes the insertion of the same segmnets of same size.)
And back to BMP. I start the BMP program with NBA=2000. The first FW I get after 2000 input records - as expected. Make checkpoint. Next FW appears after 19xx , the next after even smaller number, and so on.
For example, 100th checkpoint happens 1800 records read after 99th. But on my input data, BMP program ends successfully.
If I use NBA=8000, the behaviour is the same. Sooner or later the interval between checkpoints decreases until about 4000, and 878 abend happens.
This abend says that GetMain failed, so I decided to check what happens with CSA and ECSA. Both of them have a plenty of place - IMS have 1GB, uses 100-160MB, some times up to 600MB. ECSA has about 4500MB, and 2GB used. So, seems that no real lack of memory.