NetBackup deduplication backup jobs may fail with status 84
|Article:TECH183707|||||Created: 2012-03-13|||||Updated: 2012-03-15|||||Article URL http://www.symantec.com/docs/TECH183707|
NetBackup deduplication backups may fail because the NetBackup Deduplication Engine (spoold) is still loading fingerprints from the database into the cache. This may occur if backup jobs start running before spoold finishes loading the fingerprint cache.
Status 84 messages similar to the following appear in the NetBackup Job Details:
3/9/2012 11:57:09 AM - Critical bptm(pid=2056) image open failed: error 2060001: one or more invalid arguments
3/9/2012 11:57:09 AM - Info bptm(pid=2056) EXITING with status 84
3/9/2012 11:57:15 AM - Info bpbkar32(pid=10536) done. status: 84: media write error
NetBackup 7.5 Media Server Deduplication Pool deduplication.
When the NetBackup Deduplication Engine (spoold) starts, it reads the file segment fingerprints from the deduplication pool. Store operations (that is, writing to the disk) are disabled while the cache is incomplete. If NetBackup tries to start backup jobs while the cache is incomplete, the jobs fail.
After the spoold cache loading is complete, backup jobs and compaction jobs are enabled again. Therefore, the solution is to wait until cache loading is complete.
To determine if spoold finished loading fingerprints:
- Use your favorite tool to monitor the memory usage of services and daemons on the storage server. Wait until the memory used by spoold does not increase for 5 minutes. (When spoold loads fingerprints, it allocates memory as needed; spoold does not allocate the maximum amount of memory immediately.)
- Run the following command and examine the output for "put=Yes":
UNIX: /usr/openv/pdde/pdcr/bin/crcontrol --getmode
Windows: <install_path>\Veritas\pdde\crcontrol --getmode
"put=Yes" means that cache loading is complete. Normal operations should resume.
- If "put=No," examine the <storage_path>/log/spoold/spoold.log file for the following message:
Storage Cache Manager: load completed
The message means that spoold finished loading the fingerprints. However, the new state probably has not been pushed to all processes yet.
- If the spoold.log has the message, wait 5 minutes then run "crcontrol --getmode" again.
- If "put=Yes," cache loading is complete and normal operations will resume.
If "put=No," contact your Symantec support representative.
After the initial start of spoold after the upgrade to 7.5, spoold does not read the entire MSDP for fingerprints. Therefore, successive starts of spoold should occur much faster.
Article URL http://www.symantec.com/docs/TECH183707