Check you /etc/hosts, make sure that the host name is not on the same line with localhost. This works for me:
127.0.0.1 localhost.localdomain localhost
1.2.3.4 mycomputer.example.com mycomputer
I also have set the fully qualified hostname in /etc/sysconfig/network and in the nodes.css file.
Best I can tell, if your host name is on the localhost line, lock_gulmd tries to use the loop back device instead of the Ethernet device.
Matt
On Mon, 2005-08-29 at 19:50, yanj wrote:
Hello,
I am a beginner in GFS. I am trying to setup a GFS system based on iSCSI storage in RHEL3.
Kernel: 2.4.21-27.EL
GFS: GFS-6.0.2-25
Question
I have successful setup the GFS+iSCSI system if I use the default hostname: “localhost”.
However, if I change the hostname to another name, e.g. “test1” (update /etc/sysconfig/network and /etc/hosts), I FAIL to start lock_gulmd.
Command sequence:
Service ccsd start [ok]
Service lock_gulmd start [fail]
Service lock_gulmd status
lock_gulmd (pid 2616 2614 2611) is running...
gulm_master: gulm master not found
lock_gulmd (pid 2616 2614 2611) is running...
gulm_master: gulm master not found
I have also tried this within one machine, the error message is same.
The GFS configuration in the machine (192.168.0.22) is:
[root@test1 gfs-config]# uname -n
test1
[root@test1 gfs-config]# cat cluster.ccs
cluster {
name = "gfs"
lock_gulm {
servers = ["test1"]
heartbeat_rate = 0.9
allowed_misses = 10
}
}
[root@test1 gfs-config]# cat fence.ccs
fence_devices{
admin {
agent = "fence_manual"
}
}
[root@test1 gfs-config]# cat nodes.ccs
nodes {
test1 {
ip_interfaces {
hsi0 = "192.168.0.22"
}
fence {
human {
admin {
ipaddr = "192.168.0.22"
}
}
}
}
}
[root@test1 gfs-config]# cat /etc/hosts
# Do not remove the following line, or various programs
# that require network functionality will fail.
192.168.0.22 test1
Thanks,
Jeffrey Yan
--
Linux-cluster@xxxxxxxxxx
http://www.redhat.com/mailman/listinfo/linux-cluster
|
--
Linux-cluster@xxxxxxxxxx
http://www.redhat.com/mailman/listinfo/linux-cluster
