NetScaler HA pair periodically experiences HA failovers.

The following error is recorded in newnslog:

reltime:mili second between two records Wed Dec  2 15:26:01 2015
  Index   rtime totalcount-val      delta rate/sec symbol-name&device-no&time
      0  536581              9          1        0 ha_tot_recovers_num  Wed Dec  2 15:26:01 2015 
***ERROR*** deltatime=7 expectedtime=1449163804 actualtime=1449163664

The newnslog indicates that the NetScaler CPU tight-loop is causing the unexpected HA failover.

Complete the following steps for performance improvements:

1. Go to Shell prompt of NetScaler and run the following command:
   sysctl netscaler.ns_vpx_halt_method=2

   Perform a warm reboot for the change to take effect.
   To have this setting automatically applied every time the NetScaler virtual appliance starts, add the preceding command to the /nsconfig/nsbefore.sh file.

2. Increase the memory allocated to NetScaler VPX.

CTX121569 - NetScaler CPU tight-loop (LCT) monitor