Need to understand what’s the difference when BIND is specified on MANAGER in AGENTDEF, and decide if we need to use BIND or not.
Release : 12.0
Component : ESP WORKLOAD AUTOMATION
A socket bound to a specific local-IP address receives only packets that are targeted to that address. Outgoing packets have as their source address the address that is used to bind the socket.
BIND guarantees that all connections from the master to the agents have the same TCP/IP address.
So when you use the BIND option on MANAGER, when ESP connects to the agent, the proper TCPIP address (either the default local TCPIP or the address specified) will be used.
If you turn on INET trace, TCPTRACE will show in ESP STC output:
- When BIND is not specified on MANAGER and issue MGRADDR AGENT(agentname), following shows in the TCPTRACE:
Socket, Sock=2, ID=14843, Task=012460
Connect(1), -><ip address>:<port #>, ID=14843, Task=012460
- BIND is specified on MANAGER and issue MGRADDR AGENT(agentname), following shows in the TCPTRACE:
Socket, Sock=0, ID=17363, Task=012460
Bind, <ip address 1:*>, ID=17363, Task=012460
Connect(1), <ip address 1:*> -> <ip address 2>:<port #>, ID=17363, Task=012460
There are two differences:
1. TCPIP binds the address <ip address 1> to the socket;
2. The address <ip address 1> is on the Connect message.
IBM online doc:
ESP online doc: