This section explains the processing that HiRDB performs when an error occurs.
When a hardware or software error occurs in a HiRDB/Parallel Server, HiRDB isolates the affected units and terminates them abnormally, rather than shutting down all the units comprising the HiRDB (although there are some exceptions to this). When an error occurs in a HiRDB/Single Server, HiRDB terminates the unit, which results in complete shutdown of HiRDB, because a HiRDB/Single Server consists of only one unit.
When HiRDB has terminated abnormally and it is restarted after the cause of the error has been eliminated, HiRDB restores the system to its status immediately before the error occurred.
When HiRDB terminates abnormally, the unit will usually restart automatically, depending on the specification of the pd_mode_conf operand in the system common definition. However, if HiRDB terminates abnormally three times* in succession, the unit will not restart again automatically.
In such a case, the HiRDB administrator must eliminate the cause of the error and use the pdstart command to restart HiRDB.
* The pd_term_watch_count operand can be used to specify a maximum number of consecutive abnormal terminations that can occur. For example, if pd_term_watch_count=2 is specified, HiRDB restart will not be attempted again after two consecutive abnormal terminations.