script performance with gzip, wait and background commands

[Michael Resnick]

Hi all -

Sorry for not getting back sooner but wanted to make sure things were running smoothly. Thanks for all of your answers.

Turns out it was a memory issue, but not a memory leak. We've got a lot of things running on the machine and unknown to us, a couple of the database instances SGA sizes were increased.

We changed the dbc_max_pct kernel parm from 50 (default) to 40 and available memory seems to stay stable and enough for our processing. We've been running for almost a month without any issues.

Thanks again,

mike

[Bill Hassell]

> dbc_max_pct kernel parm from 50 (default) to 40

Unless you have a very small amount of RAM (2 GB or less), then 40% is far too large. Set the value of dbc_max_pct to use about 2-3GB of RAM, perhaps 4-6GB for 11.23 and later. An extremely large DBC is not an efficient use of RAM, especially prior to 11.23. Large SGAs that are setup efficiently will provide a bigger performance improvement than a large DBC. And if Oracle is running raw paritions rather than files, a large DBC is wasted space for Oracle.