1832678 Members
2908 Online
110043 Solutions
New Discussion

Cluster not starting

 
James George_1
Trusted Contributor

Cluster not starting

Hi

I have a 2 node cluster. After the reboot , the cluster is not coming up .. here is the error ..

#
cmruncl : Waiting for cluster to form..Jul 3 16:20:25 imcsw10 cmcld: Aborting! unexpected DLPI failure

Message from syslogd@imcsw10 at Tue Jul 3 16:20:25 2007 ...
imcsw10 cmcld: Aborting! unexpected DLPI failure

any idea ? Points assured .

rgds / James
forum is for techies .....heaven is for those who are born again !!
4 REPLIES 4
James George_1
Trusted Contributor

Re: Cluster not starting

and Here is the syslog error ..

Jul 3 16:20:25 imcsw10 cmcld: Aborting! unexpected DLPI failure
Jul 3 16:20:27 imcsw10 cmclconfd[3091]: The ServiceGuard daemon, /usr/lbin/cmcld[3092], died upon receiving the signal 6.
Jul 3 16:20:27 imcsw10 cmsrvassistd[3096]: Lost connection to the cluster daemon.
Jul 3 16:20:27 imcsw10 cmsrvassistd[3097]: Lost connection to the cluster daemon.
Jul 3 16:20:27 imcsw10 cmsrvassistd[3096]: Lost connection with ServiceGuard cluster daemon (cmcld): Software caused connect
ion abort
Jul 3 16:20:27 imcsw10 cmsrvassistd[3097]: Unable to notify ServiceGuard main daemon (cmcld): Software caused connection abo
rt


Rgds / James
forum is for techies .....heaven is for those who are born again !!
Ninad_1
Honored Contributor

Re: Cluster not starting

I believe that DLPI is something to do with Network layer two level error [ Data Link Protocol ... somrthing ]
Is there any problem with the lan connectivity for the server ? Mainly the hearbeat network ?
you can check using the command
linkloop -i
to check DLPI level connectivity for the interface on port ppa_no (which can be mapped using lanscan with the lan port you are using for connectivity) and mac_address is the address for that lan port (again can be understood from lanscan)

Are both the nodes down or the 3nd node is not able to join ?
What has changed since last running cluster?

Regards,
Ninad
James George_1
Trusted Contributor

Re: Cluster not starting

Hi

Linkloop is OK. Nothing has chnaged recently .

rgds/ James
forum is for techies .....heaven is for those who are born again !!
Prashanth.D.S
Honored Contributor

Re: Cluster not starting

Hi James,

The error occurs for two reasons:

1. All of the ports in the aggregate are coming in at different
speeds,

--AND--

2. The cluster (aggregate) is not forming with all ports.


To resolve the error, set autoport to full duplex on all ports

Best Regards,
Prashanth