Oracle Enterprise Manager Error - NMO setuid (root)

Question

I've been trying for days to fix the "Connection to host as user SYSDBA failed.ERROR: NMO not setuid-root (Unix-only)".

All, without exception, of the blogs I visited, say that in order to fix the issue, it's only needed to run the root.sh script as root, where script path is $ORACLE_HOME. I did that during Oracle Grid and Database install as requested, but re-run to make sure.

If i list the permissions of nmo and nmb binary files after running the root.sh, i get:

ls -la ./nm? -rws--x--- 1 root oinstall 34166 Apr 16 02:10 ./nmb -rws--x--- 1 root oinstall 44437 Apr 16 02:05 ./nmo

After following tutorials, i did: chown root nmo chmod 6750 nmo chown root nmb chmmod 6750 nmb

permissions went like so:

ls -la ./nm? -rwsr-s--- 1 root oinstall 34166 Apr 16 02:10 ./nmb -rwsr-s--- 1 root oinstall 44437 Apr 16 02:05 ./nmo

Above permissions are according to what most blogs said they should be like. SO I relaunched EM:

emctl stop dbconsole emctl start dbconsole

and Still got the error. I am almost performing a fresh install of Grid and Database, because I repeated the procedures multiple times, backed up /etc/oratab, deleted it and re run root.sh (so it could recreate oratab and fix possible mistakes).

I installed everything following the documentation, opened root.sh and referenced scripts so I could read them, but found nothing suspect.

My EM version is 11.2.0.3.0

I am running Oracle Linux under VirtualBox.

AGAIN: I've executed the root.sh as suggested in 99% of the websites regarding the setuid nmo error... Any clues or different procedures?

Thanks.

Answer 1

Check if the filesystem that $ORACLE_HOME is in, is mounted with the nosuid option. If it is, that will prevent any executable running as root, even if SUID bit is set.

Answer 2

Execute root.sh script in $ORACLE_HOME directory as root user:

For example:

sh /opt/oracle/product/11g/root.sh

Answer 3

Beside root.sh. You might run into my case.

Got the similar problem and found out the error came from inadequate swap space. My swap is low and run down to 300MB, this problem "Failed to launch process: Cannot run program "/opt/OEMAgent/sbin/nmo" (in directory /opt/OEMAgent/agent_inst/bin"): error=12, Not enough space" suddenly shows up.

It goes away immediately after adding additional swap space.