SAN LUN metrics relate to the request actually being serviced through read operations or write operations. General statistics contain any detailed statistics that a component collects that cannot be classified as error statistics or user-visible statistics.
| Category | Description |
|---|---|
| Total Cmds | The total number of read/write commands that are sent from any SAN host on the fabric |
| BS Requests | Any block command sent from any of the SAN hosts on the fabric |
| Non Data Xfer | Any non-block command received such as Reservation, Mode Page request, or a LUN Query |
| Non Data Xfer Response Time nS | The amount of time for the non-block command received to respond to the request. These should not hold up I/O. However, in reservation requests, I/O can be impacted because a lock could not be obtained. |
| OrderedQ | The specification from the host to send commands in a particular order |
| HeadOfQ | The specification from the host to send commands in a particular order |
| AcaQ | This is sent via IBM hosts that changes the task set with an Auto Contingent Alligence (ACA) condition |
| ACA Active | The status message that indicates whether ACA is active or not. The status message appears for IBM hosts only. |
| Busy | The Busy status is returned from the internal Block Services module in the Controller. |
| Reservation Conflict | Attempted to obtain a reservation and it was not allowed because a different host has a lock |
| Task Set Full | If the I/O request has not received a response from the internal Block Services module in the Controller within 9 seconds, the system expires and the host sends a retry command. |
| Read Before Write | Check to see if there is a Task Set Full instance, as that instance can show Read Before Write instances. If all LUNs have Disable Reference Tag Checking selected, then the Read Before Write metric is not present in the output. |
| Untagged | Queue tagging for these instances does not exist |
| SimpleQ | I/O is tagged with a simple reference |
| Check Condition Medium Error | Indicates that the system is being overloaded, or a physical error exists on the Fabric/SFP (GBIC) |
| 'Remaining' Check_Condition* | Check Conditions (SCSI STATUS) in general can cause a performance issue. So, document these instances and report the conditions to development. |