It has been reported that a NS (6.0 SP3 R13) is stuck with the Aexsvc consuming 100% of the CPU (4x) on the machine whilst logging next to nothing on the Altiris logs (NS Console load time is > 5 minutes when it doesn't just timeout).
A single NSE is being reprocessed every ~60 ms. The profiler trace gathered 100K+ events in just minutes. It looks like the 4 dispatcher threads were busy reprocessing inaccessible files, with all of them taking 60~70 ms.
Looking at the performance counters for message processing it was found some high values, in the region of 10,000 ~ 15,000 messages processed per minutes, with a total count very high.
The Client Message Dispatcher process moves the file out of the Queue onto the process folder, so failure at this stage should be caught and handled, preventing the loop from entering with an invalid filePath.
Known Issue. Few adjustments were required in our code to improve this process.
This issue was reported to the Symantec Developement Team. This issue will be addressed in a later release (post 6.0 SP3 R14)
A pointfix is available for this issue. Please see attached Pointfix_eTrack2425360_6.0_SP3_R14.zip
MINIMUM REQUIREMENT: Notification Server 6.0.6074 SP3 + R14 (See DOC5566)
HOW TO INSTALL THIS POINTFIX: 1. Extract files from the archive to the NS hard drive. Make sure PointFix is not extracted to Altiris install directory (or any subfolder to it). 2. Execute Install.cmd with administrative privileges (right-click > Run as administrator). 3. While PF installation, old files will be automatically copied to the PointFix’ subfolder ‘Backup’. 4. Close PFInstaller window.
HOW TO UNINSTALL THIS POINTFIX: 1. Make sure that Backup subfolder is located in PFinstaller’ directory 2. Execute Restore.cmd with administrative privileges (right-click > Run as administrator). 3. Assemblies will be restored from the ‘Backup’ folder 4. Close PFInstaller window.
This pointfix has the following known issues: Observed some deadlock warnings in Altiris log viewer but we can ignore them.