KBI 311700 Issue Addressed: Monitoring Task Late Or Skipped When Running Device Magic Rules For Switches With More Than 100 Ports
Version
Argent Advanced Technology 5.1A-1807-A and below
Date
Tuesday, 4 September 2018
Summary
When running Device Magic Rules against network switches with more than 100 ports, monitor tasks can be significantly late, skipped and cancelled
Further investigation shows more than 1,000 WO files with extension ‘.WR’ are accumulating in WO subfolders
The issue has been addressed in Argent Advanced Technology 5.1A-1810-A
Technical Background
Device Magic Rule is commonly used to monitor port bandwidth usage
Argent Argent Advanced Technology Engine calculates port bandwidth by polling in/out octets of each switch port and comparing with last poll values
The current values must be saved into database and used for next poll
Though saving performance value for each port into database only takes a few milliseconds, when the switch has more than 100 ports, sequentially saving the poll values into database can take significant amount of time
When there are more than 100 of such devices to monitor, it can take very long to handle all the monitoring data
As a result, the Relator cannot complete execution in time, and monitoring tasks are delayed, skipped or even cancelled
Argent Argent Advanced Technology 5.1A-1810-A addresses the issue by saving data using a thread pool
The parallel processing is the right solution for such I/O intensive work
The thread pool size is controlled by DWORD value of registry HKLM\Software\Argent\ARGENT_FOR_SNMP\UPDATE_SWITCH_INFO_THREAD_COUNT
The valid value is 0 – 100
The default value is 0, which means no thread pool is used and compatible with the old sequential processing
Resolution
Upgrade to Argent Advanced Technology 5.1A-1810-A or above
Edit value of registry HKLM\Software\Argent\ARGENT_FOR_SNMP\UPDATE_SWITCH_INFO_THREAD_COUNT to 30