Are you able to provide us the logs from the ZooKeeper instance (/var/log/zookeeper)? It should tell us why it's not starting. Please paste the logs into pastebin and provide the URL here. You just need to provide the section covering the last startup attempt and the failure
Maybe CM can't find the files, libs and so on, since the parcel-repo folder is set to non-standard path for CM (not /opt/cloudera/parcel-repo). But on the other hand this should be okay since parcel for CDH itself can be seen by CM. I really don't know.
I probably changed the owner of /opt/cloudera-manager/* directories and files from default user to "cloudera-scm". What owner should I re-set? I probably did it because of this error message you also pointed out.
For /pkg/moip/mo10755/work/mzpl/cloudera/parcels/* the owner is "root",
for /pkg/moip/mo10755/work/mzpl/cloudera/parcel-repo/* it is "cloudera-scm",
for /pkg/moip/mo10755/work/mzpl/cloudera/parcel-cache/* it is "root".
I changed the owner of all parcels folder as you said, I restarted cloudera manager server and agent too, I also changed permissions of /opt/cloudera-manager/cm-5.1.2/lib64/cmf/service/common/cloudera-config.sh:
Is /pkg/moip/mo10755/work/mzpl/cloudera/parcels or any of the directories in between on a remote host or a separate mount? Do any of the mounts have a noexec option? Run the mount command and see the flags set
Check the #! line in the script, what shell does it refer to? Is that program present on your box?
What user are you running the test as? 766 implies "other" cannot execute the script.
To verify if the cloudera-scm user is valid, run the /bin/id command on it (as root) # /bin/id cloudera-scm Does the uid and gid match that in /etc/passwd?