Through batch, the following error came up - if trying to list a program
SYSSUBST 0420 NAT0777 Buffer pool full.
coming out on the RPCS running, I could not log into production Natural using CONNECT as they same error appeared.
The way we overcame this is to recycle the natural global Buffer.
I would like to know ways to work around the error without having to recycle the natural global buffer and RPcs, can anything be done in BATCH using the Natural or adabas utilities to clear the buffer pools (force delete) or get around the error? The buffer pool space will be increased for future purposes,
When you use the GBP (or local buffer pool for that matter), there must be enough room in the pool for all active objects.
The simplest approach is as Wolfgang suggests: just make the BP bigger.
However, if you have to juggle available memory (not as often the case on mainframes now), then you can set up more smaller GBPs. For example, if the set of programs accessed from the RPC servers is typically different than that of the batch/online applications, you can also consider using different combinations of buffer pools for different areas. The RPC servers could use a local buffer pool, the batch jobs a Global BP and the onlines another Global BP.
Of course, if the different environments share a lot of modules, then a larger, common GBP makes the most sense.
A large enough GBP will improve your overall system performance as it will reduce database I/Os needed to load object modules.
