Strange. Nobody else reported that error. I need your check definition. So I can try to reproduce it.

We had the same, first of all there is a factor of * 0.01 (L.174 and L.178 in host_cpu_info.pm), which seems wrong - and even after removing that from host_cpu_info.pm the values were different from the old check_vmware_api.pl, which we now use for the CPU percentage.

So the diff/command in our case is/was:

 object CheckCommand "vmware_api_CPU_Usage" {
   import "plugin-check-command"
-  command = CustomPluginDir + "/check_vmware_esx.pl -H '$address$' -u '$vmware_user$' -p '$vmware_password$' -S cpu -s usage -w 80 -c 90"
+  command = CustomPluginDir + "/check_vmware_api.pl -H '$address$' -u '$vmware_user$' -p '$vmware_password$' -l cpu -s usage -w 80 -c 90"
 }

If i execute the check directly it also fails. With check_vmware_esx:
root@hostname:/usr/local/nagios/libexec# ./check_vmware_esx -H x.x.x.x -u nagiosuser -p passwd -S cpu -s usage -w 80 -c 90
OK: CPU usage=0.10%|'cpu_usage'=0.10%;80;90;;
With check_vmware_api:
root@hostname:/usr/local/nagios/libexec# ./check_vmware_api.pl -H x.x.x.x -u nagiosuser -p passwd -l cpu -s usage -w 80 -c 90
CHECK_VMWARE_API.PL OK - cpu usage=43.94 % | cpu_usage=43.94%;80;90

same here with 7.03

I think i tracked down what the problem can be: in L178 of host_cpu_info.pm the array entry is wrong. It should be $$values[0][3]->value. This gives me exactly the same result as check_vmware_api.pl. Shall i open a pull request for this?

edit: i forgot to mention that i tried it with both 6.5 and 7.0.3 ESXi versions.

YeThx. Faster than my answer.