Serviceguard - cmgetconf and cmquerycl giving diffrent information.

[Arun Vijay V C]

I have a service gurad cluster with two node. I was just trying to get an updated conf file from the current cluster setup.

My understanding was cmquerycl will probe and check all the hardware and configuration setup and it will create an updated file with latest information.

But in cmquerycl gives a diffrent data old data.
In cmgetconf its giving the currect conf.

#cmgetconf -v -c sm-loadbalance /home/vijayaru/cluster.ascii

omckscl2:/etc/cmcluster# cat /home/vijayaru/cluster.ascii |grep -v ^# |grep -v ^$ |more
CLUSTER_NAME sm-loadbalance
FIRST_CLUSTER_LOCK_VG /dev/vg01
NODE_NAME omckscl1
NETWORK_INTERFACE lan2
HEARTBEAT_IP 99.232.0.35
NETWORK_INTERFACE lan3
HEARTBEAT_IP 99.232.1.35
NETWORK_INTERFACE lan0
STATIONARY_IP 138.35.236.35
NETWORK_INTERFACE lan6
FIRST_CLUSTER_LOCK_PV /dev/dsk/c4t0d0
NODE_NAME omckscl2
NETWORK_INTERFACE lan2
HEARTBEAT_IP 99.232.0.36
NETWORK_INTERFACE lan3
HEARTBEAT_IP 99.232.1.36
NETWORK_INTERFACE lan0
STATIONARY_IP 138.35.236.36
NETWORK_INTERFACE lan6
FIRST_CLUSTER_LOCK_PV /dev/dsk/c4t0d0
HEARTBEAT_INTERVAL 1000000
NODE_TIMEOUT 8000000
AUTO_START_TIMEOUT 600000000
NETWORK_POLLING_INTERVAL 2000000
NETWORK_FAILURE_DETECTION INOUT
MAX_CONFIGURED_PACKAGES 150
VOLUME_GROUP /dev/vg01
Standard input: END


When I tried cmquerycl I got this,


#cmquerycl -C Newconf -n omckscl1 -n omckscl2
omckscl2:/etc/cmcluster# cat Newconf |grep -v ^# |grep -v ^$
CLUSTER_NAME cluster1
FIRST_CLUSTER_LOCK_VG
NODE_NAME omckscl1
NETWORK_INTERFACE lan0
HEARTBEAT_IP 138.35.236.35
NETWORK_INTERFACE lan6
NETWORK_INTERFACE lan2
STATIONARY_IP 99.232.0.35
NETWORK_INTERFACE lan1
NETWORK_INTERFACE lan3
STATIONARY_IP 99.232.1.35
NETWORK_INTERFACE lan7
STATIONARY_IP 99.232.14.35
FIRST_CLUSTER_LOCK_PV
NODE_NAME omckscl2
NETWORK_INTERFACE lan0
HEARTBEAT_IP 138.35.236.36
NETWORK_INTERFACE lan6
NETWORK_INTERFACE lan2
STATIONARY_IP 99.232.0.36
NETWORK_INTERFACE lan1
NETWORK_INTERFACE lan3
STATIONARY_IP 99.232.1.36
NETWORK_INTERFACE lan7
STATIONARY_IP 99.232.14.36
FIRST_CLUSTER_LOCK_PV
HEARTBEAT_INTERVAL 1000000
NODE_TIMEOUT 2000000
AUTO_START_TIMEOUT 600000000
NETWORK_POLLING_INTERVAL 2000000
NETWORK_FAILURE_DETECTION INOUT
MAX_CONFIGURED_PACKAGES 150

Any one know why cmquerycl gives wrong old data??

[Steven E. Protter]

Shalom,

The two commands are reporting on different cluster configurations. One is cluster1 the other is sm-loadbalance.

I can only conclude that one or the other is using an outdated configuration file.

SEP

[Vishu]

Hi Arun,

your both configurations are showing different clusters. but if you want to have an updated conf for the cluster, you can use cmgetconf for that. it will read the cluster binary file to fetch out the information.

whereas cmquerycl search for the nodes specified for the cluster and LVM information.

Here are the difference between two:-

cmquerycl searches all specified nodes for cluster configuration and Logical Volume Manager (LVM) information. Cluster configuration information includes network information such as LAN interface, IP addresses, bridged networks and possible heartbeat networks. LVM information includes volume group (VG) interconnection and file system mount point information.

cmgetconf obtains either the cluster configuration, not including the package configuration, or the specified package's configuration information, and writes to either the output_filename file, or to stdout. This command can be run whether the cluster is up or down. If neither cluster_name nor package_name is specified, cmgetconf will obtain the local cluster's configuration.

[Arun Vijay V C]

There is only one cluster there. Thats is sm-loadbalance. cluster1 could be the old cluster name with old configurations.

So cmquerycl gives wrong out pout here. So we can not depend on its out put for any change in the cluster.

Even the HB IP cmquerycl shows is also wrong.

So we need to take the cmgetconf output only rite?

is its always better to use cmgetconf for the cluster correct conf??

Currently the cluster having two HB IPs, and the server is having one free Lan card.

Applications needs a NIC with bonding for redundancy. But there is only one NIC free (say lanX) in the server. However, there are two NIC are use for HB IPs in the cluster.
I want to remove the second HB IP card from the cluster and create a bonding (APA) with that card and the other free card (lanX) in the server for card redundancy.

Then add lan900:1 as second HB IP card in the cluster. And Use 900:2 for application.

Out of these two HB IP I believe one is cross cable connection between the two servers and other is connected through Switch.

How can confirm this?
Which NIC is recomended to be removed from cluster for bonding? Swich connected NIC or cross cabled NIC?


Is it possible?
Could any one provide the procedure for this?

Thanks,
Arun Vijay

[Vishu]

Hi,

1. yes, always use cmgetconf for the correct cluster configuration.

2. Cross cabled Lan is for dedicated HBs and Lan from switch are providing you the external network access. your application needs the external connectivity, so better use the LAN from switch for your APA.

you can proceed as have a lan900 from apa and use it in the cluster conf and make lan900:1 for your application.

[Arun Vijay V C]

Hi Vishu,

Is it possible to use the APA NIC for cluster HB IP?

After a server reboot, while the cluster is getting started, it will look for the APA NIC lan900. By the time the whether the APA will be activated and ready for the usage?

Thanks,
Arun Vijay