Running out of OSF servers and OSFMAX is set to 30

Products

• OPS/MVS Event Management & Automation

Issue/Introduction

Currently we're running into a problem that all 30 max OSF servers had been used during some peak time we will run into problems with delayed execution of  OPS/REXX programs and problems with some other automations due to lack of OSF server.

Is that possible to increase the number of OSF servers over 30 to accommodate that or is there any other best practices for such scenarios?

Resolution

The maximum number of OSF servers is 30. However, there are other 2 types  of servers that can be used: TSL and TSP:

• OSFTSL - execute in a long running OSF server
• OSFTSP - execute in a priority OSF server

The critical processes can be segregated to use OSFTSP servers so there could be 30 dedicated servers. The other server types can handle the other requests. Also, long running programs can be dispatched to the OSFTSL servers so that they won't hold OSF servers for a long time.

The way to select each type of server is using the corresponding ADDRESS environment:

ADDRESS OSF for normal OSF servers
ADDRESS OSFTSL for long running programs
ADDRESS OSFTSP for priority programs.

Those other servers have corresponding parameters OSFTSLMAX and OSFTSPMAX which can also set to  a maximum of 30.

Increasing the value of OSFMIN may help improving performance as the overhead of adding new servers when the execution queue grows will be reduced. The same is valid for OSFTSLMIN and OSFTSPMIN.