the status of the nodes/services become concerning/bad. we found the python proces takes 100% cpu on 1 core.
after restarting the cloudera agent (on the name and datanodes) it was working again (service cloudera-scm-agent stop, service cloudera-scm-agent start).
we are running CDH 5.4.2, Parcels.
couldn't found any usefull info in the logfiles.
stopping/starting the cloudera agent service (service cloudera-scm-agent stop/start) did stop the utilisation of the cpu by the python process and the node status become "green again.
next day same issue occured.