Philpl1jb
24600 pts. | Sep 7 2009 1:36PM GMT
Not enough info
Uncommitted records are locked and a native I/O read or chain for update would encunter an error — unless they were locked by this program in this access path.
I don’t know how a Select for update would act on locked records.
So here are some questions - are you attempting to access these records in native I/O?
If by native I/O - the same program and the same access path.
Phil
As400xyz
25 pts. | Sep 8 2009 11:38AM GMT
yes…records are accessed through RPG program in INPUT mode
i.e: Using opcode READ(N)
and same access path.
As400xyz
25 pts. | Sep 9 2009 6:46AM GMT
Example:
-
PGM-A
WRITE FILE-A
WRITE FILE-B
COMMIT
-
PGM-B-
READ FILE-A
WRITE FILE-C (FROM FILE-A)
There is an error occured while writing in FILE-B. Due to this error PGM-A in message wait(MSGW) after writing a record (Ex: RCD-A) in FILE-A.
PGM-B is reading the uncommitted record (RCD-A) from FILE-A and writing into FILE-C.
In above scenario, Is there any solution to prevent reading uncommited record (RCD-A) by PGM-B? Note: PGM-A is only maintained by us.
Splat
1070 pts. | Sep 9 2009 3:51PM GMT
I’m not familiar with the particular problem, but I’d recommend putting the (E) extender on the WRITE operation (e.g.: WRITE(E)), and wrap the whole thing in a MONITOR group.






