VCS 5.0 MP1 -Solaris 10- MultiNICB and bge

Question

hello,&nbsp;I creating a cluster, at boot everything ok, when VCS starts, all the interfaces in MultiNICB resource failed !&nbsp;-node 1/etc/hostname.bge0&nbsp;&nbsp; heux167 netmask + broadcast + deprecated -failover up/etc/hostname.bge2&nbsp;&nbsp; heux167_bge2 netmask + broadcast + deprecated -failover up-node2/etc/hostname.bge0&nbsp;&nbsp; heux168 netmask + broadcast + deprecated -failover up/etc/hostname.bge2&nbsp;&nbsp; heux168_bge2 netmask + broadcast + deprecated -failover up&nbsp;configuration of MulitNICB resource in main.cf :MultiNICB MULTINICB_POSTIT_INT (&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Device @heux168 = { bge0 = "", bge2 = "" }&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Device @heux167 = { bge0 = "", bge2 = "" }&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; )if someone may help, it will be a pleasure...&nbsp;&nbsp;

bsobek · Accepted Answer

Hi Herve,&nbsp;you want to set up a link-based IPMP Configuration with VCS.&nbsp;Here is a link to the VCS-digest, witch describes your scenario.http://seer.entsupport.symantec.com/docs/317562.htm&nbsp;&nbsp;But be carefull with using link-based IPMP because there are some problems (e.g. take a look at http://sloveridge.blogspot.com/2005/09/link-based-ipmp-on-solaris-10.html&nbsp;or the "System Administration Guide: IP Services" for further information)&nbsp;RegardsBjörn

bsobek · Answer

Hi Herve,&nbsp;we have a similar configuration and it works fine. try to set&nbsp;the UseMpathd, MpathdCommand with the "-a" option (not the standard path), DefaultRouter and&nbsp; GroupName.&nbsp;MultiNICB interfaces_MultiNICB (&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; UseMpathd = 1&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; MpathdCommand = "/usr/lib/inet/in.mpathd -a"&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Device @HOST2= { bge0= "", bge1 = "" }&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Device @HOST1= { bge1 = "", bge2 = "" }&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; DefaultRouter = "&lt;IP of Default-Router&gt;"&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; GroupName = public_lan&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; )Regards,BjoernMessage Edited by bsobek on 01-21-2009 08:49 AM

herve_gregoire · Answer

thank you,&nbsp;in your configuration i think you use test addresses for IPMP.I want to have only one address per interface, and virtual IP for the applications.&nbsp;# ifconfig -abge0: flags=219040843&lt;UP,BROADCAST,RUNNING,MULTICAST,DEPRECATED,IPv4,NOFAILOVER,FAILED,CoS&gt; mtu 1500 index 2&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; inet 10.16.49.62 netmask ffffff80 broadcast 10.16.49.127&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; groupname MULTINICB_POSTIT_INT&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; ether 0:14:4f:7a:f1:d6bge2: flags=219040843&lt;UP,BROADCAST,RUNNING,MULTICAST,DEPRECATED,IPv4,NOFAILOVER,FAILED,CoS&gt; mtu 1500 index 4&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; inet 10.16.49.63 netmask ffffff80 broadcast 10.16.49.127&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; groupname MULTINICB_POSTIT_INT&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; ether 0:14:4f:7a:f1:d8

m__braun · Answer

Hi Herve,&nbsp;&gt; I want to have only one address per interface, and virtual IP for the applications.&nbsp;This seems not to be in sync with the MultiNICB Base Mode configuration requirements. See this excerpt from the Bundled Agents Guide:&nbsp;Setting up test IP addresses for Base ModeThese examples demonstrate setting up test IP addresses for your clustered systems. These IP address allows the agent determine if the NIC is working. The agent determines that the NIC is working if it receives responses for the ping packets that it sends to other nodes on the network. You do not need to performthe following steps for the floating IP addresses, as the agent takes care of this automatically.In the file /etc/hostname.qfe0, add the following two lines:north-qfe0 netmask + broadcast + deprecated -failover up \addif north netmask + broadcast + upWhere north-qfe0 is the test IP address that the agent uses to determine the state of the qfe0 network card.In the file /etc/hostname.qfe4, add the following line:north-qfe4 netmask + broadcast + deprecated -failover upWhere north-qfe4 is the test IP address that the agent uses to determine the state of the qfe4 network card. In the above example, north-qfe0 and north-qfe4 are host names that correspond to test IP addresses. north is the host name that corresponds to the test IP address.&nbsp;&nbsp;Regards&nbsp;Manuel

herve_gregoire · Answer

Thanks to all,&nbsp;i had some doubts about the solution but it works immediately.&nbsp;Enjoy with Symantec products !

Forum Discussion

VCS 5.0 MP1 -Solaris 10- MultiNICB and bge

5 Replies

Related Content

Re: Solaris11 Project Tuning (Shared Memory) for Clustered Netbackup

Solaris 10 + Zones + ZFS + VCS

IPMP Issue with MultiNICB Agent on VCS 5.0 MP1

Solaris 11.4 support

Re: Sync users on cluster nodes and solaris zones?

Recent Discussions

Configure two Mount type resources of nfs FStype attribute using the same share

order

key registration and reservation

Verifying that primary and dr clusters replication is synced

vcs can create logical nic