How to diagnose problems with Optimized Latency Mode (OLM)

Resolving The Problem

Perform these steps:

• Display the OSA-Express3, OSA-Express4S, or OSA-Express5S by using D TCPIP,NETSTAT,DEVLINKS. Ensure OptLatencyMode displays as Yes.
• If OptLatencyMode displays as No, make sure OLM is configured on the INTERFACE statement.
• If OLM is already enabled,
• look at the console for either of these messages:
• EZD0045I INTERFACE interface_name DOES NOT SUPPORT OLM
• EZD0046I ERROR error_code ENABLING OLM FOR INTERFACE interface_name
• If you see EZD0045I, and you are using OSA-Express3, issue a D vtamproc,TRL,TRLE= or D TCPIP,,OSAINFO command and verify the OSA Code Level. Look at the 2097DEVICE or 2098DEVICE Preventive Service Planning (PSP) buckets to see which level is required for OLM, and make sure that the displayed LIC level is at least an OLM level. All OSA code levels for OSA-Express4S and OSA-Express5S support OLM.
• If you see EZD0046I with an error_code of 0002, ensure there are no more than 4 concurrent interfaces to this OSA-Express3, OSA-Express4S, or OSA-Express5S. If you receive this error code when enabling an OLM interface, it means that this is the FIRST connection on the target port to attempt OLM enablement, but the maximum number of non-OLM connections has already been established on the target port. Reduce the total number of connections on the port to less than the maximum or do not enable OLM.
• If OptLatencyMode is Yes on the NETSTAT display, enable tuning statistics for the OSA-Express3, OSA-Express4S, or OSA-Express5S by using the F vtamproc,TNSTAT,TRLE= command.
• With traffic flowing over this OSA-Express3, OSA-Express4S, or OSA-Express5S, look for non-zero counts in the following messages:
• IST2316I EARLYINO = earlyino EARLYINT = earlyint
• IST2317I ULPRETUO = ulpretuo ULPRETU = ulpretu
• If you see non-zero counts displayed in those messages, then OLM is operating.
• If you do not see non-zero counts displayed in those messages, verify you have either enabled WLMPRIORITYQ or have used SETSUBNETPRIOTOSMASK to ensure that traffic is directed to queues 1, 2 or 3.