Intermittent connectivity issues between random agent and AutoSys scheduler

Products

• Autosys Workload Automation

Issue/Introduction

The Autosys Scheduler is experiencing intermittent connectivity issues with some agents on remote job execution servers. Jobs become stuck in STARTING status when the agent fails to respond to AUTO_PING checks, requiring manual intervention via FORCE_STARTJOB and causing significant execution delays.

Cause

In many Linux/UNIX environments, the default "open files" limit is set to 1024. For a production Autosys Scheduler or Application Server, this is often insufficient due to the number of concurrent database connections, log file handles, and network sockets required. For a busy Linux production environment, the default ulimit (often 1024) is generally too low, as every network socket and opened file consumes a file descriptor.  

Resolution

The System Administrator needs to increase the ulimit on the machine(s) that are imapcted.

Note: The max number of files that can be opened depends on available RAM, as each descriptor consumes a small amount of kernel memory.

 

 

Additional Information

For more information regarding the ulimit parameter, please reference the appopriate vendor documentation for your OS.