ProcessMetricsCollector doesn't print 'command not found' error correctly

[paisleyrob]

I've created a CentOS 6.6 chroot with nginx-1.6.2/passenger-4.0.59 for my rails application. Inside this chroot is very sparse (69 packages total) such that /bin/ps does not exist.

When I start nginx the following gets spammed into the error.log:

[ 2015-03-02 21:17:16.9083 5513/f776fb70 Pool2/Pool.h:760 ]: ERROR: The '-opid,ppid,%cpu,rss,vsize,pgid,uid,command' command failed
Backtrace:
in 'long long unsigned int Passenger::ApplicationPool2::Pool::realCollectAnalytics()' (Pool.h:867)
in 'static void Passenger::ApplicationPool2::Pool::collectAnalystics(Passenger::ApplicationPool2::PoolPtr)' (Pool.h:754)

It looks to me like ProcessMetricsCollector v4.0.59 uses ps under the hood and blows up repeatedly.

If /bin/ps is required (appears it is), I expected to see a /bin/ps: File Not Found type of error and possibly exiting.