cmclconfd and unknown source of syslog error

[Peter A. Berger Jr.]

I cannot seem to find this error message anywhere in ITRC/Google and cannot decifer what it means. Any help would be great. Thanks.
"cmclconfd[15878]: Unable to get local username
cmclconfd[15878]: Unable to initialize communication
inetd[781]: hacl-cfg/udp: Exit status 1
inetd[25675]: getpwnam: root: No such user"

The only command running at this time every morning is a script:"
. /apps/oracle/local/efector_env smi1
PROGLOC=${ORACLE_BASE}/local/dbr
export PROGLOC
/usr/bin/date >>${PROGLOC}/DBR.log
${ORACLE_HOME}/bin/sqlplus -s / @${PROGLOC}/DBR_Mail >>${PROGLOC}/DBR.log"

[Michael Steele_2 · Accepted solution]

Check /etc/rc.config.d/cmcluster and make sure AUTOSTART_CMCLD = 0. Unless you're in a ServiceGuard environment then it should be 1.

If yes to ServiceGuard then check all nodes for duplicate /etc/inetd files with similar entries like the following:

hacl-cfg dgram udp wait root /usr/lbin/cmclconfd cmclconfd -p
hacl-cfg stream tcp nowait root /usr/lbin/cmclconfd cmclconfd -c
hacl-probe stream tcp nowait

Chances are root is misconfigured on the node in question.

[Jeff Schussele]

Hi Peter,

If you do indeed have Service Guard, then make sure you have a cmclnodelist file in /etc/cmcluster & that it contains the following syntax

hostname1 root
hostname2 root

for all nodes in the cluster.

Also make sure the command or script you reference in your post is not trying to do any of the following commands if you don't have Service Guard - cmquerycl, cmviewcl, cmcheckconf or cmapplyconf.

HTH,
Jeff

[melvyn burnard]

Well I think here is your clue:
inetd[25675]: getpwnam: root: No such user"

This indicates you are using NIS, and it has not been able to resolve the root user id.
If not, hten you ar using soe other method of "synchronizing" login_id/password. If you cannot resolve the root user, then the Sg daemons, such as cmclconfd (which use root) will fail.
Does this happen on just one node? or all the cluster nodes?

[Peter A. Berger Jr.]

New insight on this error. (I'm such a dumbass UNIX newbie) This script was run by "efdba" at 4am. Also at 4am "root" was copying /etc/passwd.bak to /etc/passwd
The reason why there is no /etc/passwd before 4am is that we rename /etc/passwd to /etc/passwd at 2100hrs ET each night to disable users from logging into the machine. I'm assuming that we'd get the "no user found" error message because
1. There's no passwd file and
2.the sqlplus -s command is referencing an OS user.
Sorry to trouble y'all, but your answers are helpful anyways.

[melvyn burnard]

You have it 100% correct, it will not find the valid userid

[Pete Randall]

Rather than renaming /etc/passwd, you might want to consider using the nologin functionality of "security" (see man security):

NOLOGIN This parameter controls whether non-root login can be disabled by the /etc/nologin file.

NOLOGIN=0 Ignore the /etc/nologin file and do not exit if the /etc/nologin file exists.

NOLOGIN=1 Display the contents of the /etc/nologin file and exit if the /etc/nologin file exists.

Default value: NOLOGIN=0


Pete

[Jeff Schussele]

No reason to have to manipulate the PW file to disable logins - just do this

1) Ensure the file /etc/default/security contains NOLOGIN=1
2) Place a file called /etc/nologin on the system with whatever text you want to display to the users, such as "No logins allowed at this time."

Then when you wish to allow logins remove the /etc/nologin file. This can be easily scripted & you won't run into the above error & this has no effect on valid root logins.

HTH,
Jeff