Current status rebuild

When a restarting member has completed its own current status rebuild (CSR), it checks and waits for every other member to finish CSR. If non-starting members exist, peer CSR is performed.

During current status rebuild, the following tasks are accomplished:
  • The SCA is rebuilt from the log by reading it forward from the last checkpoint.

    All restarting members add entries to the indoubt transaction ID (XID) list in the SCA from information contained in their logs. If an indoubt XID entry cannot be added to the SCA, the member abnormally terminates with reason code 00F70606.

  • Db2 determines all outstanding units of recovery (URs) that were interrupted by the previous termination.
  • If the lock structure is lost, all partition and page set P-locks are reacquired by reading information from the log. These locks are retained locks until the end of restart.

During CSR, you see messages similar to the following messages. (The phrase in parentheses is not part of the output.)

 DSNR001I @DB3ADB2 RESTART INITIATED
 DSNR003I @DB3ADB2 RESTART...PRIOR CHECKPOINT RBA=00000000DC4E
 DSNR004I @DB3ADB2 RESTART...UR STATUS COUNTS
 IN COMMIT=0, INDOUBT=0, INFLIGHT=0, IN ABORT=0
 (End of current status rebuild for member DB3A)
 
 DSNR021I @DB3ADB2 DSNRRGRC DB2 SUBSYSTEM MUST PERFORM
 GROUP RESTART FOR PEER MEMBERS