Steve, First let me say that I have OCFS 1.0.13 (and OCFS 1.0.12) with 9205. But I have been exactly in your position...it was a while back but let me try. [ btw, I found OCFS GUI not very useful nor a *complete* utility to the command line options...still Oracle Support pushed me to use it...I did not...I gave up on it.] >>I finally got the old nodes to go away but >>I had to reformat the volume I used for ocfs. Yes with no where to turn and no help, saddly I (incorrectly) did the same thing. In my case I think we changed (interconnect) IP's (after a fried mother board with embed NIC) which is also a big "no, no" with OCFS...and we had problems with duplicate "node numbers" (as seen in) "tuneocfs" output because we "regen" new config file. Our solution was to; explicitly reclaim the old node number by specifying "-o reclaimid" in the mount command as follows: mount -t ocfs -o reclaimid /dev/sda1 /o01 However in your case you could have simply removed the old/unwanted entries shown in tuneocfs output (for each disk/mount device); ---tuneocfs -l /dev/sdc1 tuneocfs -l /dev/sdX ==> lists all the nodes configured tuneocfs -N <nodenum> /dev/sdX ==> clears the appropriate entry. You can delete the spurious entry using tuneocfs like so: "tuneocfs -N 1 /dev/sdc1" [root@db01 root]# tuneocfs -l /dev/sdc1 # Name IP Address IP Port Node GUID = ================================ =============== ======= ================================ 0 db01 192.168.12.40 7000 621F2EEDBB65697D877F000F1F6EE4EE 1 db02 192.168.12.41 7000 A2790979698C1CBA42F7000F1F6EE56D [root@db01 root]# tuneocfs -N db01 /dev/sdd1 Clearing node number 0 used by node db01 Proceed (y/N): y Changes written to disk. I found this to be helpful; "OCFS_Users_Guide.doc" Also I made myself some personal notes I titled; OCFS & Network Interface Card - ocfs_uid_gen - tuneocfs.txt OCFS Commands.txt I will send them to you off line. hth Chris Marquez Oracle DBA -----Original Message----- From: oracle-l-bounce@xxxxxxxxxxxxx on behalf of Steve Perry Sent: Tue 11/22/2005 5:33 PM To: peters@xxxxxxxxx; _oracle_L_list Subject: Re: Reconfigure OCFS 1.0.14 Hi Peter, It was ocfs1. I finally got the old nodes to go away but I had to reformat the volume I used for ocfs. Once I did that, I could reconfigure it. I'm glad I didn't have anything on it and could afford to do that. I opened a ticket with Oracle. Of course they responded after I resolved it, but they said that Oracle doesn't support renaming a host after it's been installed. I'm not sure why it would be so tightly integrated into the host so a rename would be such a big deal. I'll check out the list also. Thanks, Steve ----- Original Message ----- From: "Peter Sylvester" <peters@xxxxxxxxx> To: "_oracle_L_list" <oracle-l@xxxxxxxxxxxxx> Sent: Tuesday, November 22, 2005 1:19 PM Subject: Re: Reconfigure OCFS 1.0.14 -----Original Message----- From: oracle-l-bounce@xxxxxxxxxxxxx on behalf of Steve Perry Sent: Mon 11/21/2005 10:54 PM To: oracle-l@xxxxxxxxxxxxx Subject: Reconfigure OCFS 1.0.14 I hope somebody can help with this one, but I wasted a whole day trying to figure it out. I installed ocfs 1.0.14 on RHEL3. I was ready to install the CRS for RAC 10g r2 and found out the hostname had to be changed for our standards. I changed it and rebooted the server. I started ocfstool and saw that the "configured nodes" were still the old server names. I renamed the ocfs.conf file, rebooted and started ocfstool and it still showed the old hostname. I generated new keys thinking that would help, but now I have 4 configured nodes (old and new hostnames for each ip address). How does ocfs determine what nodes are configured and how do I remove nodes? Will if fix it, If I uninstall ocfs packages and reinstall them? I'm not sure I need to worry about it, but I didn't want to get deeper in the config and have it cause problems. The most frustrating part is the documentation doesn't mention it. I looked in var/log/messages and tried google and metalink, but came up empty. thanks, Steve