Symantec NetBackup KM for PATROL

Release Notes for v2.8.03

Home  Previous  Next

What's New

Job Configuration: Child jobs, typically retry jobs, are now created under their parent jobs with the job status alerts snoozed by default. This reduces the amount of events triggered for a job failure. It is also possible to trigger alerts for the child job failures by changing the settings in the job configuration menu.
New Report: The Available Media report is added to Volume Pools to report on all available media.
NBUDatabaseStatus: New parameter added to check the availability of NetBackup database (NBDB).
Inactive standard parameters labelled <collector>ExecTime were added to monitor the performance of each collector. Please refer to the Reference Guide for detailed information.
A 'no command execution window' can be defined for the KM to avoid running NetBackup commands during regular maintenance window everyday. Please refer the Troubleshooting section of this guide for detailed information.
Three new parameters added to the NBU_JOB_CONTAINER application class:
NBUJobActiveBackupReduction displays the reduction in number of active backup jobs (NBUJobActiveBackupCount) between collection cycles.
NBUJobQueuedBackupCount displays the number of backup jobs in 'Queued' state.
NBUJobQueuedBackupReduction displays the reduction in number of queued backup jobs (NBUJobQueuedBackupCount) between collection cycles.

Changes and Improvements

Erroneous Exit Status filter: the filter is changed to support both failure and erroneous filters. They are used for setting NBUJobState to Failed (10) or Error (6) state respectively.
NBUDatabaseCollector: The collector now searches all backup jobs in the NetBackup database to determine the last database backup and set NBUDatabaseElapsed. Previous versions were searching only in the new backup jobs since the last job collection cycle. The alert conditions on NBUDatabaseElapsed are also turned back on to trigger alerts if there are no database backups found or if they are too old.
Two new states added to parameter NBUPolicyState:
Inactive policy is monitored by setting NBUPolicyState to “Inactive”.
Next backup of each active policy is calculated and if jobs were not started as per the schedule, NBUPolicyState will be set to “Not Started”.
Two new states added to parameter NBURDriveState to trigger alarms through NBURDriveStatus:
When robotic drives change to AVR mode, they will be alerted by setting NBURDriveState to “Invalid”
If a previously discovered robotic drive is missing, NBURDiveState will be set to “Missing”.
All configuration menu (KM commands) support regular expressions. For example, the Policy configuration menu can be used for disabling all “test” policies, which will also stop monitoring their jobs.
Backup job report is stored and shared between collectors to avoid rerunning bpdbjobs command. This reduces the number of commands executed by the job collector and improves the performance considerably, especially at large NetBackup sites.
Robotic libraries and drives discovery commands are shared and optimized for performance.
Collectors are synchronized to reduce the load on the agent while using the data as they become available.
Unnecessary debug messages are removed.
Faster KM failover in clustered NetBackup environment. The server discovery interval is reduced from 5 minutes to 40 seconds, which speeds up detecting server level changes as well as clustered NetBackup server failover.
The KM Status report now provides collector/discovery collection time information as well as additional performance details. This report is stored under PATROL_HOME/lib/NBU/km_status.txt on all managed nodes.
NBUJobCollector has been significantly optimized to perform job collection more rapidly.

Fixed Issues

On Unix platforms, daemon collector command & arguments are modified to return only the relevant daemon lines from the process list. This corrects an error in setting the parameters for Process Count, Memory Size, CPU Utilization & CPU Duration.
Incorrect border alarm on NBULibraryMediaScratchCount triggered alerts when there are more 100 scratch media. The border alarms are now removed to stop the inappropriate event. Also the border alert on NBULibraryMediaAvailablePercent is removed to stop alerting when the value is unknown.
An error in a script made the KM fail to calculate the scratch count affecting NBULibraryScratchCount & NBULibraryScratchPercent parameters.
Duplicate client instances were created when clients were configured with short hostname and fully qualified hostnames. The client collector now takes the first occurrence of such clients and ignores the other duplicates.
In clustered NetBackup systems, passive nodes run as media server and change to master during failover. The KM did not detect this node type change without an Agent restart. Node changes are now applied during the next collection cycle and all collectors are refreshed immediately.
Error reading “Failed to obtain lock on SOWLogWriteLock” was repeatedly displayed in the System Output Window and NBU.log on Windows managed nodes. This error is now corrected and the log file NBU.log is rotated successfully.
When disabling monitored components using the KM configuration menus it disables all instances with the same label, but under different parent. This affects policy clients, robotic drives, disk pools & volumes.
If any of the commands used by the KM failed to return a required field, the KM failed to process the remaining fields in the command output. The missing field are now ignored.
Configuration menu changes at the container level did not take effect without an Agent restart.
When upgrading KM from v2.7.x to v2.8.x, some of the instance level KM configuration data stored in the PATROL configuration were not upgraded.
In clustered NetBackup servers, when a cluster node failed to report the cluster status, the KM changed to “Temporary Single-node Mode” without checking the KM status on the other cluster nodes. The KM now checks the remote node status prior to changing monitoring mode.
Backup Job Report: Long command output lines prevented the report to be properly processed. The script has been modified to handle long command lines.
The syntax error in affecting the drive stats collection has been fixed.