Neoview Database Administrator's Guide (R2.2)
Readers :
---------
1. /opt/testdata/WISC-INSERT
Writers :
---------
1. 16.107.129.58:1138
2. 16.107.129.58:5063
3. 16.107.129.58:9052
4. 16.107.129.58:13053
5. 16.107.129.58:17038
Cumulative statistics from previous runs:
Requested Applied Rejected Affected
30200 30200 278 29522
----------------------------------------------------------------
[16:29:20] MASTER> MASTER Execution Started.
[16:29:22] MASTER> INFO : WRT_TARGET ODBC/MX FastPath has been enabled.
[16:29:22] MASTER> INFO : INIT_TCGS Wating for intialization threads to complete.
.
.
.
The task instance log for the recovery run is written to the same task instance log file as that
created during the initial run of the task instance. Cumulative record counts are reported for all
the previous runs of the task instance. Bad records are also written to the same .bad files that
were created for the initial run of the task instance.
For more details, see “Logs” (page 115).
Reconfiguring a Recovery-Enabled Task
If a recovery-enabled load task fails, you cannot change the configuration for that task before
re-running the task to a successful completion. The only exception is that you can change the
recovery flag to N, to disable recovery, then rerun the task. This runs the task from the beginning
and the recovery information is ignored.
When the task instance is re-run with recovery disabled, the recovery information from the
previous run on the Neoview system is deleted. The .rcvy file for the previous run on the Linux
platform is also deleted.
CAUTION: Use caution when disabling recovery for a task, because this removes any chance
of recovering the task in the case of failure.
To run a task with recovery disabled, set the recovery flag, -rf, to N when you configure the
load task.
This is an example of a task configuration request with recovery disabled:
gcmd -rt CONFIG_TASK_LOAD \
-tk REG-1E-1 -rf N \
-it FILE \
-ii /opt/tandem/genus/mbhatt/testdata/WISC-INSERT \
-ot TABLE -oi DCCAT.DCSCH.WISC_NAR \
-un dbmgr30 -pw dbmgr30 \
-of Y
Duplicate Records in .bad Files
When you run a recovered task, it is possible for duplicate records to be inserted into .bad files
on the Linux loader platform. This can only happen if:
• The gelserver data-load process terminates abnormally, and
• The process had just finished writing a rejected record to the input.bad file, and
• The process terminated before the .rcvy file could be updated
Recovery 119