/usr/bin/collector-runner WebServiceMonitor wakes up every 10 seconds and does a search looking for files in /data/project/*/service.manifest
tools-service-02 generates around 400-500 NFS ops/sec. We'd rather avoid that load on our NFS server, considering the performance problems we experience.
It seems the reaction time of 10 seconds might afford some tweaking to better match what our environment can really offer.
There's additional load on LDAP services but I haven't measured those. Network throughput from LDAP requests is higher than NFS, so changing the wake up time would also help with that.