Parameter settings for the AMCC adapters for the Sun host system

5: Default.

When the AMCC adapter or driver detects that the fibre-channel link is up (and there is no I/O activity), it sends a test frame (or heartbeat) to itself to verify link integrity. The test frame is sent at the interval specified by this parameter. If the test frame does not complete, there might be a link problem. In this situation, the driver initiates error recovery to reestablish a good link. A value of 0 disables the heartbeat.

1000: Default.

Delay (msec) after the link is up before port discovery begins, allowing the link to stabilize and protecting against a possible I/O surge. This timer is reset every time the link comes up. The default value is adequate for most configurations.

5000: Default.

Delay (msec) before retrying after receipt of an I/O operation with a SCSI Busy status from a target. The number of retries is based on the Solaris retry count associated with the I/O operation.

30:

Delay (seconds) before failing all I/O operations for an offline target. If the delay timer expires, all I/O operations for the failed target is returned to the application. A zero value disables failover.

0: False.

Boolean parameter for enabling SCSI target resets for timed-out I/O operations. When the timer expires (usually 60 seconds, as specified by the upper layers), the driver issues a target reset to attempt to clear the device (which might be either too busy to respond or stuck).

5: Default.

Number of times an I/O operation is retried due to receipt of a SCSI queue full status from a target. The delay between retries is based on the QfullRetryDelay parameter.

5000: Default.

Delay (msec) before retrying after receipt of an I/O operation with a SCSI queue full status from a target. The number of retries is based on the QfullRetryCount parameter.

50: Default.

Sets the LUN I/O recovery interval (in msec) after the driver reconnects to a disk. It is a global parameter affecting all targets, and determines how long the driver waits after a port is discovered until sending I/O operations to that port. Some devices might require more time to flush an I/O operation that was in progress prior to a link going down; if this is the case, increase the value of this parameter.

3: Default.

5: Default.

Delay (seconds) after driver creation to allow the network to stabilize, discover ports, and build the driver database. Increase this value if targets are being discovered too late in the restart process.

3: Default.

Total number of fabric login (FLOGI) attempts before giving up logging in to a switch. Failure prevents participation on a fabric topology.

10: Default.

Specifies the amount of time (in seconds) that the driver waits for a fabric login (FLOGI) to be accepted. The value should be increased only if the switch to which the host adapter is connected requires more time to respond to a FLOGI. The number of retries is configured with the FlogiRetryCount parameter.

5: Default.

Total number of port login (PLOGI) attempts before failing to log in to a SCSI target.

30: Default.

Defines the number of seconds that the driver waits for a successful port login (PLOGI) attempt. The maximum number of attempts is defined by the PlogiRetryCount parameter. Some devices might take longer to respond to PLOGIs; if this is the case, increase the value of this parameter.

1789: Default.

Total number of concurrent exchanges (also called transfer control blocks) that the adapter allows. To optimize performance, set this parameter to match the memory capacity of the hardware.

Recommended setting: 0.

Function: Flag for enabling target offline when target port fails.

1: Default. (Enable).