Nonstop Database, HiRDB Version 9 System Operation Guide

[Contents][Index][Back][Next]

1.6.3 Notes on HiRDB startup processing errors (applicable to HiRDB parallel server configurations only)

This section provides notes on errors during unit startup.

Organization of this subsection
(1) Handling system log files
(2) Starting the dictionary server

(1) Handling system log files

Conditions
Either condition 2 or 3 must be satisfied.
  1. Multiple servers are defined at the unit whose startup processing failed.
  2. The startup mode is normal startup or restart after planned termination.
  3. The startup mode is restart after forced termination or abnormal termination, and pd_log_rerun_swap=Y is specified.

Event
When a unit is started, HiRDB starts all servers at the unit concurrently. If an invalid system definition operand is detected or an error occurs in one of the servers, startup of the entire unit fails. In such a case, the following events might occur:
  • At a server whose startup processing is completed, allocation of the current system log file is completed (the KFPS01221-I message is output)
  • At a server whose startup processing is not completed, the current system log file is not allocated (the KFPS01221-I message is not output)

Action
In such a case, HiRDB allocates a new current file the next time unit startup is performed. The current file allocated during the previous unit startup is closed (it is not reused). The HiRDB administrator must take one of the actions listed in the table below to reallocate this closed system log file as the current file.
(a) Not reinitializing the system log file

After executing pdlogunld or pdlogchg for the corresponding system log file, one of the actions listed below can be taken, depending on HiRDB's activity status:

HiRDB's activity status HiRDB administrator's action
During HiRDB operation Execute the pdlogopen command on the corresponding system log file.
Before normal startup of unit No action is required. This file is identified as the available current system log file during unit startup.
Before unit restart After the unit is restarted, execute the pdlogopen command on the corresponding system log file.
(b) Reinitializing the system log file

Procedure
  1. Execute the pdlogunld or pdlogchg command on the corresponding system log file:
    pdlogunld -d sys -s b001 -g syslogfgp03 -o /unld/unldlog01
  2. Use the pdlogrm command to delete the corresponding system log file:
    pdlogrm -d sys -s b001 -f /unt1/sysfile01/log01
  3. Use the pdloginit command to initialize the corresponding system log file:
    pdloginit -d sys -s b001 -f /unt1/sysfile01/log01 -n 5000

(2) Starting the dictionary server

Event
If you perform any of the following operations, and then start a unit, the dictionary server acquires the RDAREA configuration information:
  • Normal termination of BES
  • Database structure modification utility (pdmod)
    [Figure] RDAREA addition (create rdarea)
    [Figure] RDAREA expansion (expand rdarea)
    [Figure] RDAREA reinitialization with the with reconstruction option specified (initialize rdarea)
    [Figure] RDAREA attribute alteration (alter rdarea)
    [Figure] RDAREA move (move rdarea)
    [Figure] RDAREA replica definition (replicate rdarea)
    [Figure] RDAREA configuration information copy (define copy rdarea)
    [Figure] RDAREA unification (recast rdarea)
  • Replica status switching of a replica RDAREA (pddbchg)
  • Switching of the current RDAREA after online reorganization (pdorchg)
  • Reflection processing after online reorganization (pdorend)
If the dictionary server has not started, the system retries acquiring RDAREA configuration information up to 60 times. If the information cannot be acquired after 60 retries, the acquisition process terminates and startup processing of the BES unit fails.

Action
Start the dictionary server, and then start the BES unit.