A checkpoint dump file is used to collect table information required for OpenTP1 recovery.

In order to handle possible errors, OpenTP1 collects checkpoint dumps in multiple checkpoint dump files, using the round-robin scheduling method. The most recent checkpoint dump generation is required for full recovery of OpenTP1. During full recovery, all recovery journals collected from the time the most recent checkpoint dump generation was collected until the time OpenTP1 terminated are also required.

If the most recent checkpoint dump generation cannot be read, the immediately preceding checkpoint dump generation is used. If this checkpoint dump cannot be read, the generation collected immediately before it is used. Thus, the checkpoint dump files are searched backward for a usable generation, and all recovery journals collected after collection of the checkpoint dump to be read are required. If needed journals have been lost (because they have been overwritten by other data), OpenTP1 cannot be recovered. Currently, two generations are guaranteed. The journals collected after the most recent checkpoint dump are guaranteed.

The following figure shows the sequence in which checkpoint dump files are used.

Figure 4-8 Sequence of using checkpoint dump files

OpenTP1 collects checkpoint dumps for a server at the following times:

• When server startup or restart processing is completed
• When system journal files are swapped
• When the number of journals specified in the system journal service definition have been collected since the previous checkpoint dump collection
• When server termination preparations are completed

OpenTP1 manages a file group of checkpoint dump files in the online mode in the statuses listed below.

• Overwritable status or writing status
  A checkpoint dump file is overwritable when it does not contain a checkpoint dump needed for OpenTP1 full recovery or checkpoint dump collection is currently underway.
  
• Non-overwritable status
  A checkpoint dump file is overwrite-protected when it contains a checkpoint dump needed for OpenTP1 full recovery.
  
• Reserve
  A checkpoint dump file is on reserve status when it is closed and cannot be used in the online mode without first being opened.
  

When OpenTP1 is started, all file groups specified in the checkpoint dump service definition for which ONL has been specified are opened. The opened file groups are placed on overwritable status. File groups that cannot be opened or for which ONL is not specified are placed on reserve status.

By using the jnladdpf command, a new physical file can be allocated online to a checkpoint dump file for which a file group has been specified in the checkpoint dump service definition but a corresponding physical file has not (that is, with the definition command jnladdfg specified but jnladdpf not specified). The file thus allocated is in reserved state; it can be used online by opening it with the jnlopnfg command.

When checkpoint dump collection occurs, an output destination is selected from the overwritable file groups, and the selected file group is placed on writing status.

When checkpoint dump collection is completed, the file group on writing status is placed on non-overwritable status, and the oldest non-overwritable file group is placed on overwritable status.

When OpenTP1 is started, the guaranteed number of generations plus at least one is required for each file group (except reserve file groups). If use of the fall-back facility is specified in the fall-back operation option of the checkpoint dump service definition (jnl_reduced_mode=1 or 2), at least two more unreserved file groups than the guaranteed number of generations are needed. If there is no overwritable file group, OpenTP1 issues an error message to the message log file and terminates abnormally.