Workload not distributed as expected

Use the following checklist to diagnose workload distribution problems:

Verify that the Advisor's update interval value is not inordinately large. The Advisor must wait at least two update intervals before beginning to receive enough data to properly calculate weights when an application becomes available or when an Agent is started.
Allow at least three update intervals to expire after an application is started before re-examining the distribution of workload. If workload is occasionally being sent to overloaded applications or systems, adjust the update_interval downward so workload distribution can react more quickly to the pace of new workload requests.
Periodically issue display commands on the Advisor to check the weights of members within the group in question. Determine whether the weights are consistent with the expected behavior. If the weights are not consistent with expected behavior, see z/OS Communications Server: IP System Administrator's Commands for more information about how to analyze the member weights; if all releases in the sysplex are not V1R9 or above, note the restrictions and limitations described in this section. If these are consistent with the expected behavior, investigate the problem at the LB. For more information, see z/OS Communications Server: New Function Summary.
Verify that the Advisor's agent_id_list value contains the IP addresses and ports that each Agent is bound to on the MVS™ systems where the application exists, if not using AT-TLS. If it does not match and you are not using AT-TLS, correct the agent_id_list statement on the Advisor or the advisor_id statement on the Agent.
If using AT-TLS with SERVAUTH access control checks to validate connections between the Advisor, Agents, and external load balancers, see Diagnosing Application Transparent Transport Layer Security (AT-TLS). In addition, ensure that the SERVAUTH class is active. Ensure that the EZB.LBA.LBACCESS.sysname.tcpsysplexgroupname resource profile is defined and that the user ID associated with the external load balancer has READ access to it. Ensure that the EZB.LBA.AGENTACCESS.sysname.tcpsysplexgroupname resource profile is defined and that the Agents have READ access to it. On the system console where the Advisor is running, look for message EZD1280I which indicates that a connection attempt using AT-TLS failed. This message has specific reason codes which indicate the reason for the failure.
Issue display commands at the Advisor to make sure that members of the group in question are not unexpectedly quiesced or unexpectedly unavailable (AVAIL status is NO).
If AVAIL is NO because sysplex problem detection and recovery issued message EZD1973E then the TCP/IP stack must either be restarted or at least 10 minutes must pass from the last occurrence of the problem (for example, an abend). See the Sysplex problem detection and recovery section in z/OS Communications Server: IP Configuration Guide for more information.
Issue display commands at the Advisor for all system-level members in the sysplex to verify that the MVS systems have the expected residual capacity.
Check the log file for ERROR or WARNING messages and take the appropriate corrective action. If ERROR and WARNING level log messages are not enabled, enable them and recheck the log file later.