(a) Physical file re-creation

If an error occurs while a checkpoint dump is being written, OpenTP1 outputs a message to the message log file, closes the erroneous file group, changes it to a reserve file group, reselects the output destination from the overwritable file groups, and outputs the checkpoint dump again.

If the corresponding file group is not placed on reserve status, the erroneous file group must be closed and changed with the jnlclsfg command to a reserve file group.

The file group placed on reserve status by the jnlclsfg command can be changed to an overwritable file group by the following procedure:

1. Delete the file group with the jnlrm command.
2. Re-create the file group with the jnlinit command.
3. Open the created file group with the jnlopnfg command.

(b) Physical file reallocation

If an error occurs on the physical file allocated with the jnladdpf command in the online mode, another physical file is allocated.

For dual-system operation, another file can be allocated by manipulating only either system A or system B. Note that the file group is temporarily placed in reserved status. Therefore, the contents of both of the physical files are invalidated. The file group cannot be used if either of the systems is disconnected.

The following shows the re-allocation procedure:

1. Close the file group where the error occurred.
   Use the jnlclsfg command to close the file group and place it in reserved status.
   
2. Delete the physical file.
   Use the jnldelpf command to delete the physical file where the error occurred. The system (A or B) can be specified.
   
3. Allocate another physical file.
   Use the jnladdpf command to allocate a normal physical file. The system (A or B) can be specified.
   
4. Open the file group.
   Use the jnlopnfg command to open the file group and make it overwritable.
   

(c) Re-create the system where the error occurred

When one-system operation is available in a dual system configuration, if an error occurs on either of the systems for the checkpoint dump file, the system where the error occurred can be re-created to make both systems usable. This system re-creation can be executed only when the file group is overwritable.

The following shows the re-creation procedure:

1. Close the system where the error occurred.
   Use the jnlclsfg command to close the system where the error occurred. If this command is executed when the system has already been closed, a command error occurs.
   Do not execute jnlclsfg to a normal system. If executed, the contents of the physical file become invalid.
   
2. Delete the system where the error occurred.
   Use the jnlrm command to delete the physical file of the system where the error occurred.
   
3. Re-create the physical file.
   Use the jnlinit command to re-initialize the physical file to be allocated to the system where the error occurred.
   
4. Open the physical file.
   Use the jnlopnfg command to open the re-created physical file.
   

When OpenTP1 is restarted, it opens all the file groups that were open during the previous OpenTP1.

(a) Non-duplicated checkpoint dump file configuration

Any file group that could not be opened or that was on reserve status during the previous online operation becomes a reserve file group.

(b) Duplicated checkpoint dump file configuration

(i) Either system A or B file failed to open

If one-system operation is available, continue processing by opening the normal system. If one-system operation is unavailable, the file group is placed in reserved status after the read is completed or when the file group becomes overwritable.

(ii) Both systems A and B files opened successfully

Recover the system by reading from system A.

If the read from system A fails when one-system operation is available, system B is used. If the read from system A fails when one-system operation is unavailable, the data is read from system B and after that (when the file group is non-overwritable, after it becomes overwritable), the file group is placed in reserved status.

If the checkpoint dump file required for recovery cannot be read due to an open error, etc., OpenTP1 outputs a message to the message log file and continues reading as much of the checkpoint dump file as possible.

No addition, change, or deletion can be made to the checkpoint dump service definition during restart.