server built from ignite image from SG cluster

[Andy Macleod]

I've build a server from a ignite image from a node in a service guard cluster.

Just wanting to check how to disable it service guard.

I've set the 0 in the /etc/rc.config.d/cmcluster file, but i'm still have strange happenings on my lan cards.

When the server comes up, you can connect for a min or so on the NIC but then the card goes down and no traffic is seen coming out of the server.
Do people think this is service guard issue or have i got another problem???

[RAC_1]

ps -ef|grep -i cmcl will tell you if any of the cluster related processes are running or not.

you may have problem with dead gateway probe setting.
ndd -get /dev/ip ip_ire_gr_probe
ndd -h ip_ire_gw_probe for help. If set to 0.

Anil

[A. Clay Stephenson]

If you have set AUTOSTART_CMCLD=0 in /etc/rc.config.d/cmcluster AND have rebooted AND have not manually executed cmrunnode then MC/SG is not at play. You can check with "ps -ef | grep cmcld". At the very least, your stationary IP should be fine.

One thing that immediately occurs to me is that you now have duplicate IP addresses. Have you changed the IP address of this host so that it does not conflict with the original?

[Andy Macleod]

Checked for cmcl processes - none found.
So I'm thinking not SG related.

As for changing the ip, yup it was done also the server was put on a completly different network.

I'm now onto thinking duplex settings with the switch, my GB nic's are set to auto neg and cant be nailed to 1000FD just wondering if the switch is set to 1000fd and my nic have dropped to 10hd or something.

[Patrick Wallek]

If you are using Gb networking, then I believe BOTH the switch AND the NIC have to be set to AUTO-NEG for it to work correctly.

[Steven E. Protter]

You definitely have an IP address conflict.

The Ignite Interface lets you intervene at the console by keyboard and change that. Its right on the networking tab. You might want to re-image the server.

Go to the console of the new machine.

/sbin/init.d/net stop

modify all the ip addresses in /etc/rc.config.d/netconf

/sbin/init.d/net start

Better yet, boot the after making your changes.

Long term, if this box is not intended to be part of the SG cluster, than you may want to make a golden image that doesn't include SG.

SEP

[Stuart Abramson]

I can't think of a reason why this should happen. Here are some things I can think of:

1. Wipe out /etc/cmcluster. If the config files are gone, then you can't have any MC/SG action taking place.

2. Disable cluster start on boot (already stated above):
cd /etc/rc.config.d/
vi cmcluster
set:
AUTOSTART_CMCLD=0

3. set automatic activation of VGs to 1 on both nodes. This is usually set to 0 for MC/SG hosts:
cd /etc
cp lvmrc lvmrc.old
vi lvmrc
set
AUTO_VG_ACTIVATE=1

(This will prevent non vg00 VGs from being started on boot if set to 0.)

[Sridhar Bhaskarla]

Hi,

Most likely it's not Serviceguard issue since you disabled cmcluster file. Look at each of your /etc/rc.config.d/hp*conf files and see if you have MAC addresses hardcoded in there. If so, remove them and reboot the system. Otherwise, they will cause MAC address conflicts.

You can also get more clues by looking at the nettl logs. Run 'netfmt /var/adm/nettl.LOG00|LOG01'.

-Sri

[Nguyen Anh Tien]

1. In order to disable cluster start on boot, do as following
vi/etc/rc.config.d/cmcluster
edit AUTOSTART_CMCLD=0
2. Check the health of Lan card
[trn0:/etc/rc.config.d] lanscan
Hardware Station Crd Hdw Net-Interface NM MAC HP-DLPI DLPI
Path Address In# State NamePPA ID Type Support Mjr#
0/0/0/0 0x00306E5CAFCB 0 UP lan0 snap0 1 ETHER Yes 119
0/10/0/0 0x00306E9469C3 3 UP lan3 snap3 2 ETHER Yes 119
[trn0:/etc/rc.config.d] linkloop 0x00306E5CAFCB
Link connectivity to LAN station: 0x00306E5CAFCB
-- OK
[trn0:/etc/rc.config.d] linkloop 0x00306E9469C3
Link connectivity to LAN station: 0x00306E9469C3
-- OK
[trn0:/etc/rc.config.d]
Note: You should check Hdw state column to detect whether Lan up or down. If linkloop fail you should check Lan cable,
network performance,
3. Check IP for each interface:
[trn0:/etc/rc.config.d] arp -a |awk '{print $2}'|sed s/://g|sort

[Andy Macleod]

Fixed it but I'm not too happy with the config.

My netstat -rn is
Dest/Netmask Gateway Flags Refs Use Interface Pmtu
127.0.0.1 127.0.0.1 UH 0 2247 lo0 4136
192.168.50.67 192.168.50.67 UH 0 1240 lan1 4136
192.168.50.64 192.168.50.67 U 2 0 lan1 1500
127.0.0.0 127.0.0.1 U 0 0 lo0 4136
default 192.168.50.65 UG 0 0 lan1

My netconf file is

ROUTE_GATEWAY[1]=192.168.50.65
ROUTE_DESTINATION[1]=default
ROUTE_COUNT[1]=1
ROUTE_MASK[1]=255.255.255.192
ROUTE_ARGS[1]=""
DHCP_ENABLE[1]=0
IP_ADDRESS[1]=192.168.50.67
SUBNET_MASK[1]=255.255.255.192
INTERFACE_NAME[1]=lan1
BROADCAST_ADDRESS[1]=192.168.50.127
INTERFACE_STATE[1]=up

My Lanscan is
Hardware Station Crd Hdw Net-Interface NM MAC HP-DLPI DLPI
Path Address In# State NamePPA ID Type Support Mjr#
0/0/0/0 0x000D9DAE0DC6 0 UP lan0 snap0 1 ETHER Yes 119
0/8/0/0 0x00306EF47121 1 UP lan1 snap1 2 ETHER Yes 119
0/9/0/0 0x00306EF461F3 2 UP lan2 snap2 3 ETHER Yes 119
0/10/0/0 0x00306EF4712A 3 UP lan3 snap3 4 ETHER Yes 119
0/12/0/0 0x00306EF47120 17 UP lan17 snap17 5 ETHER Yes 119

And it Didnt work, kept dropping out after a few secnds after a reboot.

I did set_parms initial which added the following to the netconf file.

ROUTE_GATEWAY[0]=192.168.50.65
ROUTE_COUNT[0]=1
ROUTE_DESTINATION[0]=default

And it Works.

My issue is I have 2 defaults in my netconf file (i think this is a bad thing)

Can anyone explain?