topic Re: cluster problem in hp-ux in Operating System - HP-UX

cluster problem in hp-ux

ramikhlb — Sat, 06 Nov 2010 07:46:01 GMT

hello,
i have 2 rx4440 running hp-ux we had a power failure EVA and servers was down.
The problem is when the servers start all the vg on the storage deactivated I change the file and put the auto =1
Now all vg activated even after reboot
But in bdf I can see only vg00 , I mount all the logical volume on one nod ( it give me error that the volume is corrupted and need check ) I used fsck and the problem solved
On the second node I cannot mount the logical volume it gives me the same error of corrupted logical volume and fsck didnâ t solve the problem on the second node

Ioscan â fnC disk is showing all the disks in eva ( noconnection problem)

Cluster is giving the error
Cannot view the cluster configuration: No such file or directory.
Either this node is not configured in a cluster, user doesn't have
access to view the cluster configuration, or there is some obstacle
to viewing the configuration. Check the syslog file for more information.
For a list of possible causes, see the Serviceguard manual for cmviewcl.

attached the syslog of the 2 nodes and the pkgs log ( the power failure was in 23 oct)

your assist will be appreciated

Re: cluster problem in hp-ux

jskatiyar — Sat, 06 Nov 2010 16:23:37 GMT

As per node LUO logs cmclconfd can't read cluster binary file. Run cmviewconf OR "cmscancl -n LUO" to confirm binary is in good shape. Also check rng module is loaded in kernel as syslog indicates. I believe system have multinode package. However, you need to check ioscan claims device c8t0d1 and other disks (run ioscan -funCdisk) as package logs shows PV does not exists. Check if due to EVA side changes device paths have been changed. Next is to run fsck on one node and do not activate VG on other node at the moment.
Keep /etc/lvmrc AUTO_VG_ACTIVATE=0 and try to start cluster.
Never initiate cluster with command like "cmruncl -n tchicapa". It will become cause of split brain syndrome. For more details read public document "emr_na-c02067755" available at Business Support Centre.

Re: cluster problem in hp-ux

ramikhlb — Sat, 06 Nov 2010 20:22:00 GMT

hello jskatiyar,
thank you for your reply
the run ioscan -funCdisk show all disk as claimed and the c8t0d1 is claimed too

cmscancl -n luo
can not remsh to the system luo

cmviewconf
Unable to open communications to configuration daemon: I/O error
Unable to connect to configuration database.
cmviewconf: Either binary file does not exist, or the user doesn't
have access to view the cluster configuration.

Also check rng module is loaded in kernel as syslog indicates

sorry i didn't understand, how to check it ?

i can see that the binary file is not exist or corrupted , any idea how to fix this problem ? is it the file in /etc/cmcluster ?

i'm new in hp-ux and your help is appreciated

Re: cluster problem in hp-ux

jskatiyar — Sun, 07 Nov 2010 08:57:23 GMT

It is turning into critical state. "Unable to open communication to configuration daemon and database" error means you need to check /etc/cmcluster/cmclconfig does exist in both the nodes. If it is misplaced at one node copy it into same directory from other node. Check hacl* lines are not commented into /etc/inetd.conf file and /var/adm/inetd.sec is not blocking communication for cmcl* daemons those run at 53** port
To check rng module status run "kcmodule -v rng" command.
If problem does not solve by these, time is to open a case with hp support.

Re: cluster problem in hp-ux

ramikhlb — Sun, 07 Nov 2010 10:16:59 GMT

hello

LUO:[/]#kcmodule -v rng
Module rng (0.1.0)
Description Strong Random Number Generator
State unused (because of an error)
State at Next Boot loaded (as requested)
Capable loaded unused
Depends On interface HPUX_11_23:1.0

Check hacl* lines are not commented into /etc/inetd.conf file

no it is not commented

/var/adm/inetd.sec is not blocking communication for cmcl* daemons those run at 53** port

how to know if it is bloked or not ? please gibe me the steps

configuration file on both node was each one with differant name but same content so i change the name to be the same in both node , this was done before i asked for help and not now

any idea??
if i have an ignite tape and i make restor the problem will solved ?
i tryied to make ignite but when i interrupt the auto boot
hpux>
i enter boot p3 to boot p8 with no success and the tape is shown on ioscan
any idea ?

thanks for your help

Re: cluster problem in hp-ux

DeafFrog — Mon, 08 Nov 2010 08:30:54 GMT

hi ,

what does (pls paste o/p)
cmgetconf cluster.ascii
cmcheckconf -C cluster.ascii

Regards,