Troubleshooting
Problem
This document is a short host server customization update describing how to set a subnet mask for clients where only the final octal of the TCP/IP address varies.
Resolving The Problem
The subnet mask for host server customization entries must align correctly with the specific TCP/IP addresses to be routed. As a rule, the following is true:
If the final octal of the TCP/IP address is the only portion that varies between a set of customized clients, the subnet mask is defined as 255.255.255.255. If this change is not made, once the TCP/IP address is filtered through the subnet mask. The individualizing portion of the address is gone and the host server routing code has no way to determine which client IP is routed to which subsystem.
Example:
Assume the following two TCP/IP addresses are to be customized: 192.168.5.3 and 192.168.5.4. The subnet mask is defined for both as 255.255.255.0 (which is the default value before version 6.1). 192.168.5.3 is defined to run in the CLIENTONE subsystem, and it is to be rejected when this subsystem is unavailable. 192.168.5.4 is defined to run in the CLIENTTWO subsystem, and it is to be rejected when this subsystem is unavailable. All other clients are to route to the QUSRWRK subsystem.
On a specific connection, after the mask is applied to each specific TCP/IP address, the results are the same for both: 192.168.5.0. When 192.168.5.4 connects to the IBM System i system, the mask is applied and results in 192.168.5.0. The connection is routed to the first entry that is valid for this value. In this case, it would be routed to the ClientOne subsystem rather than the ClientTwo subsystem.
To avoid this inconsistent behavior, the subnet mask for this type of client entry is changed to 255.255.255.255. Once this correction is made and the customized host server is restarted, the TCP/IP addresses will retain their differentiating final octal values after the subnet mask is applied. The connections are routed correctly to the defined subsystems.
The order that entries are added to the list also affects the outcome. The following rules must be kept in mind when you set up a customization:
If the final octal of the TCP/IP address is the only portion that varies between a set of customized clients, the subnet mask is defined as 255.255.255.255. If this change is not made, once the TCP/IP address is filtered through the subnet mask. The individualizing portion of the address is gone and the host server routing code has no way to determine which client IP is routed to which subsystem.
Example:
Assume the following two TCP/IP addresses are to be customized: 192.168.5.3 and 192.168.5.4. The subnet mask is defined for both as 255.255.255.0 (which is the default value before version 6.1). 192.168.5.3 is defined to run in the CLIENTONE subsystem, and it is to be rejected when this subsystem is unavailable. 192.168.5.4 is defined to run in the CLIENTTWO subsystem, and it is to be rejected when this subsystem is unavailable. All other clients are to route to the QUSRWRK subsystem.
On a specific connection, after the mask is applied to each specific TCP/IP address, the results are the same for both: 192.168.5.0. When 192.168.5.4 connects to the IBM System i system, the mask is applied and results in 192.168.5.0. The connection is routed to the first entry that is valid for this value. In this case, it would be routed to the ClientOne subsystem rather than the ClientTwo subsystem.
To avoid this inconsistent behavior, the subnet mask for this type of client entry is changed to 255.255.255.255. Once this correction is made and the customized host server is restarted, the TCP/IP addresses will retain their differentiating final octal values after the subnet mask is applied. The connections are routed correctly to the defined subsystems.
The order that entries are added to the list also affects the outcome. The following rules must be kept in mind when you set up a customization:
- Add entries for specific, single IP addresses before you add IP address ranges.
- Insert entries for single IP addresses in ascending order of IP address then add IP address range entries in ascending order of the first IP address in the range.
[{"Product":{"code":"SWG60","label":"IBM i"},"Business Unit":{"code":"BU058","label":"IBM Infrastructure w\/TPS"},"Component":"Host Servers","Platform":[{"code":"PF012","label":"IBM i"}],"Version":"Version Independent","Edition":"","Line of Business":{"code":"LOB57","label":"Power"}}]
Historical Number
355207973
Was this topic helpful?
Document Information
Modified date:
25 June 2021
UID
nas8N1015862