Server Tuning Guide – Known Issues Related to VM Performance and Tuning

Delays in resource allocation are known to cause two different types of issues. These can be specific to individual sessions or affect the performance of an entire machine during peak operating times.

JobYields – These interruptions in service to individual sessions leave log entry that mentions JobYields in the StayLinkedStdErr.log file. This issue causes individual session threads to become unresponsive until the affected sessions are terminated. This issue is resolved on Windows and Linux Server Versions 15.1 and newer which default to the OpenJ9 JVM instead of the prior HotSpot runtime.

Note that this may also take place on hardware systems that are underpowered or have bursts of activity by other processes that delay resource availability to StayLinked.

Full Server Hesitation or Delays – This issue typically affects virtualization cluster systems during shift changes or high demand periods. It is most often caused when the virtualization system attempts to dynamically move or re-appropriate resources, causing delays in resource availability to the StayLinked server process. Clustered VMs might be reviewed for this if you experience delays across all sessions that eventually clear. StayLinked Server version 14.5 introduced enterprise performance improvements that improve the startup and/or shutdown of sessions in large numbers over a short time period. If your Server shows delays across all sessions at the same time, it may be important to upgrade to a current server release.

The best solution is the configuration of your virtualization tool to guarantee adequate resources are provided to every process that is considered mission-critical to your business.