Recovering and job accounting

 

If a job ends abnormally, the final accounting entry is written and all previously written accounting entries appear in the journal. If an abnormal system ending occurs, the following accounting data is lost to the last routing step or last end-of-accounting segment, whichever occurred most recently.

• Information on the number of lines and pages printed

• Number of files created

• Database put, get, and update operations

• Communications read and write operations

• Auxiliary I/O operations

• Transaction time

• Number of transaction fields

• Time active

• Time suspended

After an abnormal system ending, the job completion time in the journal will not be the same as that in the CPF1164 message. The message uses the time nearest to that of the system ending, but the job accounting journal entries are sent to the journal during IPL, and the job completion time is the current system time, which is later than the time when the abnormal system ending occurred.

If the system ends abnormally, some journal entries can be lost. These are the entries that are written to the journal but not forced to disk (this is equal to FORCE(*NO) on the Send Journal Entry (SNDJRNE) command). They include the following:

• JB entries caused by a Change Accounting Code (CHGACGCDE) command

• DP and SP entries

Whenever a job completes, the last accounting code entry is forced to disk (as if FORCE(*YES) were specified on the SNDJRNE command). Whenever an accounting entry is forced to disk, all earlier entries in the journal, regardless of which job produced them, are forced to disk.

 

Exception

If only *PRINT accounting is specified on the system, there will not be any job ending FORCE(*YES) journal entries done. therefore, if a critical accounting entry is written by a CHGACGCDE command you want to ensure it will not be lost in case of an abnormal system ending, you can issue a SNDJRNE command and specify the FORCE(*YES) option. If files are also to be journaled to the accounting journal, any database changes are always forced to the journal, and this causes all earlier accounting entries to also be forced.

If an abnormal system ending occurs or you change an accounting code of a job other than your own, the qualified job name in the first 30 bytes of the JARES field in the journal entry describe the system job that wrote the JB entry at the next IPL and not the job that used the resources. The JAJOB, JAUSER, and JANBR fields should be used for analysis purposes.

• Damaged job accounting journal or journal receiver
  If damage occurs to the journal or to its current receiver so that the accounting entries cannot be journaled, a CPF1302 message is sent to the QSYSOPR message queue, and the accounting data is written to the QHST log in the CPF1303 message. The job trying to send the journal entry continues normally. Recovery from a damaged journal or journal receiver is the same as for other journals.

• Accessing the CPF1303 Message
  To access information from the CPF1303 message, create a high-level language program.

 

Parent topic:

Managing job accounting