sg cluster of 2 HP Integrity VM's

[joseph pareti]

I keep getting an "access denied" error when trying to reach the quorum server (version a.03.00.00 (3)

sg 11.18

i have 2 cluster nodes and 1 quorum server:
#asosw078.deu.hp.com 16.58.249.143 (node)
#asosw111.deu.hp.com 16.58.249.176 (node)
#tcosw020.deu.hp.com 16.58.249.234 (qs)

# cmcheckconf -C $SGCONF/J_M_cluster.config -v
...
Unable to probe cluster configuration. Unable to connect to quorum server at tcosw020 from node asosw078: Permission denied.
Check the syslog file for more information.
Unable to probe cluster configuration. Unable to connect to quorum server at tcosw020 from node asosw111: Permission denied.


#tail /var/adm/syslog/syslog.log
...
Dec 18 13:18:02 asosw111 cmclconfd[6804]: Version reply error from quorum server at 16.58.249.234: access_denied.
Dec 18 13:18:02 asosw111 cmclconfd[6804]: Unable to probe quorum server 16.58.249.234. Permission denied.


I have set up .rhosts on all nodes and on the qs, so that remsh works in any direction. I have also distributed the ssh keys so that root can login from any of the 3 hosts to any other host w/o password prompt.

I am also including the cluster config file

[melvyn burnard · Accepted solution]

have you set up the Quorum Server with it's authorization file?
It has to have this, it does NOT need .rhosts to work

[joseph pareti]

yes I did
#cat /etc/cmcluster/qs_authfile
heartbeat1.abc.de.efg
tcosw017.deu.hp.com
heartbeat2.abc.de.efg
tcosw018.deu.hp.com
asosw111
asosw078

I am not sure if there is a version skew, the qs is on hp-ux v3, the 2 cluster nodes are on hp-ux v2.

[melvyn burnard]

and what does the qs log say?

[Torsten.]

For Compatibility see the "Quorum Server Compatibility Matrix"

http://docs.hp.com/en/6202/QSSupportMatrix.pdf

[joseph pareti]

# tail /var/adm/qs/qs.log
---

Dec 18 13:09:54:0:Access denied to host: 16.58.249.143
Dec 18 13:09:54:0:Access denied to host: 16.58.249.176
Dec 18 14:00:15:0:Access denied to host: 16.58.249.143
Dec 18 14:00:15:0:Access denied to host: 16.58.249.176
Dec 18 14:20:17:0:Request for lock /sg/cluster1 succeeded. New lock owners: tcosw017,tcosw018
root@tcosw020:/root>

[melvyn burnard]

well that tells me that the QS is working correctly, and that it is unable to rexolve the two hostnames you have added in.
when you added them in to the qs_auth file, did you restart the QS?
and can the QS server correctly reslove those hostnames to ip addresses abnd vice versa?

[joseph pareti]

>did you restart the qs

no, i didn't; how can I do it? there is an entry in /etc/inittab that restarts it if it dies, but i see no "qs" process on the system

root@tcosw020:/root> grep qs /etc/inittab
qs:345:respawn:/usr/lbin/qs >> /var/adm/qs/qs.log 2>&1
root@tcosw020:/root>

[melvyn burnard]

Well asuming you are in runst a 3,4 or 5, it should be running, and your log file indicates it is:
Dec 18 14:20:17:0:Request for lock /sg/cluster1 succeeded. New lock owners: tcosw017,tcosw018


Do ps -ef |grep qs to check this.
To restart it, either kill the resultant PID from the previous step, or try doing init q

[joseph pareti]

indeed, killing the processes "qc" has solved the problem, the server is respawned and apparently has read the modified config file. Thanks to all of you.

[melvyn burnard]

glad you sorted it out.
If yoyu are happy that all is fixed, please close this thread by clicking teh Close Thread button

[joseph pareti]

Many thanks to everyone.