Video Screencast Help

VCS INFO V-16-20002-211 Netlsnr :monitor:Monitor procedure /opt/VRT Sagents/ha/bin/Netlsnr/LsnrTest.pl returned the output: su: Unknown id: oracle

Created: 13 Feb 2013 • Updated: 21 Apr 2013 | 6 comments
This issue has been solved. See solution.

Hello,

I am new to veritas. We got this error on our server and the listener failed over to passive node. There was no network issue that time also the sudo was working perfectly. Not sure what went wrong. Any help on this error is much appreciated.

2013/01/13 03:25:23 VCS INFO V-16-20002-211 Netlsnr :monitor:Monitor procedure /opt/VRT

Sagents/ha/bin/Netlsnr/LsnrTest.pl returned the output: su: Unknown id: oracle

 

Jithu

Discussion Filed Under:

Comments 6 CommentsJump to latest comment

mikebounds's picture

Is oracle a local user or an ldap or NIS user or was anyone running any commands at the time that change the order of whether system uses files of NIS.

Mike

UK Symantec Consultant in VCS, GCO, SF, VVR, VxAT on Solaris, AIX, HP-ux, Linux & Windows

If this post has helped you, please vote or mark as solution

Jithu's picture

yes, oracle is a local user. And there was no change done to nsswitch.conf file. The version of veritas was 5.1. let me know if you need more details.,

Jithu

Satish K. Pagare's picture

What is the operating system you are running VCS and Oracle on? Has this error happened anytime again or earlier? Also could you please send the package details for VRTSvcsea package? 

This error means, the agent during its detail monitor checks for the Oracle Listener component failed to do the checks using lsnrctl due to the "user id not found error". 

mikebounds's picture

VCS is reporting error with su command, so unless VCS is passing an invalid username, the issue is with the su command, so you should perhaps contact your O/S vendor.  The only possibilty I can see in it being VCS is if the agent is running something like su "oracle " with a space or other unseen character in the user name, but you can look at the code (in LsnrTest.pl) yourself and it is just running "su $Owner" where Owner is passed as an argument to the perl script (unless you are using a custom LsnrTest.pl) and if VCS were passing a space in Owner arg, then it is very unlikely it would do this just once as oppose to all the time, so the agent would never work.

Mike

UK Symantec Consultant in VCS, GCO, SF, VVR, VxAT on Solaris, AIX, HP-ux, Linux & Windows

If this post has helped you, please vote or mark as solution

Jithu's picture

Thanks Mike&Satish..

 

Operating system is Solaris 10 and veritas is 5.1

Powerbroker is installed on this server and it is controlling sudo access. We have checked with powerbroker team and they informed there was no issue at that time. Also we haven't faced this issue on other veritas cluster nodes running on solaris 10. 

Jithu

mikebounds's picture

I would think it is most likely the issue was caused by Powerbroker if this interferes with su.  The only way I could see this being VCS, is as I said earlier, if VCS were passing a space or other unseen character in Owner arg

You could place some debug in the script so if this happens again you can find out what issue is, by amending end of the LsnrTest.pl file something like:

 

if ($? != 0) {
        alarm 0;
        print $LsnrResult;
        ### NEW LINES
        open (ERRFILE, '>>/var/tmp/lsnr.err');
printf ERRFILE "ZZ%sZZ\n", $Owner;
close (ERRFILE);
### END OF NEW LINES

exit $VCSAG_RES_OFFLINE ;#resource is offline } else { alarm 0; exit $VCSAG_RES_ONLINE; # resource is online }

 
This should produce a file like:
ZZoracleZZ
 
So if instead it looks like:
ZZoracle ZZ
 
then you know VCS is issue, else the issue is with su.
 
Note it is valid to use your own LsnrTest.pl for the MonScript attribute, so there is no issue editting this file.
 
Mike

 

UK Symantec Consultant in VCS, GCO, SF, VVR, VxAT on Solaris, AIX, HP-ux, Linux & Windows

If this post has helped you, please vote or mark as solution

SOLUTION