Fujitsu M2488 manual Error Recovery Procedures, Edrc Error Recovery, Retry Methods

MAINTENANCE AND SERVICING M2488 PRODUCT GUIDE

8-6 ERROR RECOVERY PROCEDURES

The IDENTIFY message is used by the initiator or target to establish the physical path between the ini- tiator and target for a specific logical unit. Bit 6 of the IDENTIFY message is defined as follows:

Bit 6: When the initiator sets this bit to a one, the initiator allows disconnect. If the initiator sets this bit to a zero, the initiator does not allow disconnect. When the target performs a disconnect/reconnect sequence, an implied RESTORE POINTERS operation is performed by the initiator. This results in all data for the command being retransmitted from the initiator unless the target has issued a SAVE DATA POINTER message prior to disconnection.

The target always sets bit 6 to zero during reconnection.

If an error occurs during data transfer, the target is automatically restored to the beginning of data.

8-6.1 EDRC Error Recovery

Certain errors during EDRC operation are recoverable. This category of errors occurs during data transfer and requires the data to be re-transmitted in order to record the data on tape. Included in this category are the following related errors:

Initiator record expands past 451 KB during compaction process.

Initiator record expanded past the available buffer space.

Hardware error occurred. (e.g. timeouts, compression/decompression errors, etc.)

8-6.2 Retry Methods

There are four retry methods the target is capable of using to recover the above errors:

1.Automatic retry from the internal retry buffer for blocks up to 64 kB (65,536 bytes)

2.Explicit Restore Pointers

3.Disconnect/reconnect with an implied Restore Pointers

4.Aborted Command

The above methods are tried in order as determined by the initiator (MESSAGE REJECT message or IDENTIFY message).

The error recovery for the first method is performed internally within the target. This retry method can be disabled via feature mode 1, bit 0 (0x01), using the CHANGE DEFINITION command, or from the operator panel SETTING menu, option 77:S.FT1 (refer to the User’s Guide, Chapter 4).

The error recovery for the second and third methods is completed when the host restores the data pointers and retransmits the data.

The error recovery following an aborted command is completed when the host reissues the WRITE command for the failing data.

For all four retry methods, the target performs the first retry in the same data compression mode as the original data transfer. For any additional retries, the target automatically switches to reblocking in non-compacted mode. The target automatically switches back to the mode defined prior to the error after the retry is successful or after the third retry has failed and a permanent error has been reported to the initiator.

If the initiator does not support the restore pointers mechanism for error recovery, the initiator must reject the RESTORE POINTERS message with the MESSAGE REJECT message.

If the initiator does not support the disconnect/reconnect mechanism for error recovery, the initiator must not allow the target to disconnect via the IDENTIFY message.

If the initiator does not support both the RESTORE POINTERS message and disconnect/reconnect, the target can be configured to not perform these retry methods. The target is configured by setting

8-30 CG00000-011503 REV. A April 1997