[Linux-cluster] The case and the question ?
Yazan Al-Sheyyab
yazan at ccs.com.jo
Tue Jun 21 08:30:19 UTC 2005
Hello,
Here is the case and what done till now :
i have installed RHEL_ES_V3_update 4 on the two nodes and the shared can be seen from the two nodes through /dev/cciss/c0d0 i had configured each node and done with the packages needed by oracle, then
i had made two partitions on the shared each with 100M for using as Raw1 and Raw2 for the cluster suite quorum then i had add them on the /etc/sysconfig/rawdevices file then i run service rawdevices restart command and everything run ok, then i had installed RedHat cluster suite update 4 on the two nodes and configured the raws needed there . THEN and after that i had mailed the list about the need of using the cluster suite to have a cluster running oracle RAC 10g Then the list has answered my question that i dont need the cluster suite.
So , i have continued and installed the RedHat Global File System (GFS 6 ) update 4 on the two nodes and i have used the documentation in configuring it .
i have made another two partition on the shared for use for Oracle , and i have build the configuration files for GFS and after that and before starting Oracle installation i tried to test shutting down one of the nodes after mounting the two gfs partitions on the shared , then an error message appeared about the Lock_gulm server the it is still running and it still apears for a long time without shutting down even i power them down from the power botton on the server .
i dont understand the lock_gulm or the files of ccs sonfiguration files cause i maybe bulid an error in any of them (i think the fence file cause i didnot understand it ) .
i will give you the files that i have build till now , and im using two proliant ML250 servers and a shared of raid 5 disks remaing a 208.5 G of its space ...... here is the files :
the pool00.cfg for the first partition on the shared (for using oracle) :
poolname pool00
minor 0 subpools 1
subpool 0 128 1 gfs_data
pooldevice 0 0 /dev/cciss/c0d0p5
the pool01.cfg for the second partition on the shared (for using oracle) :
poolname pool01
minor 1 subpools 1
subpool 0 128 1 gfs_data
pooldevice 0 0 /dev/cciss/c0d0p6
then i had done with the pool_tool command for them and everything is ok.
then i had create a directory for the ccs files on the home of the root :
/root/clu
and i have put in the /root/clu the following ccs files :
cluster.ccs :
cluster {
name = "oracluster"
lock_gulm {
servers = ["orat1", "orat2"]
heartbeat_rate = 0.3
allowed_misses = 1
}
}
fence.ccs :
fence_devices {
admin {
agent = "fence_manual"
}
}
------------------------------------------------------------
nodes.ccs :
nodes {
orat1 {
ip_interfaces {
eth1 = "10.0.0.2"
}
fence {
human {
admin {
ipaddr = "10.0.0.2"
}
}
}
}
orat2 {
ip_interfaces {
eth1 = "10.0.0.3"
}
fence {
human {
admin {
ipaddr = "10.0.0.2"
}
}
}
}
}
--------------------------------------------------------------------------
here is the files ... i had done with them . but i dont under stand the fence file use and is it true for my case i had choosed to be a manual fence , but do i need another hardware to use for it or what ???????
Sorry for the long E-mail but just to let the reader have info about my case correctlly.
any answer for my case please . ????????????????/
Thanks
Regards
-------------------------------------------------
Yazan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20050621/9e09e770/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tech.gif
Type: image/gif
Size: 862 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20050621/9e09e770/attachment.gif>
More information about the Linux-cluster
mailing list