VG with a two nodes cluster with Oracle RAC

[jorge palma]

I can not activate 4 V.Groups that ServiceGuard (with RAC extension) is using for a Oracle RAC.

Oracle RAC it is installed and configured with SLVM + Raw devices.

In node1, i can see this VG activated, available and shared. Everything it's OK. But in node2, i can not activate VG with the needed configuration. When i try activate them i get the error:

agnesi/#vgchange -c y -S y /dev/vgpro02data
vgchange: Failed to set the specified attributes for volume group "/dev/vgpro02data".

I atach screen captures about SMH on both servers.

How could i activate VG for node2.

Thanks in advance.

[Steven E. Protter]

Shalom,

Activate the volume group in shared mode.

The volume group must be on shared storage, visible to both systems.

SEP

[sujit kumar singh]

Hi,


I would like the RAC experts to correct me as i suggest the following.

if the VG is activated already on one node as an SLVM vg then it has the SHARED and Cluster aware attributes set.

On the other node you need to activate in the shsred mode as

#vgchange -a s /dev/vgprod2data.


please wait for some more suggestions from RAC experts.

regards
sujit

[Sajjad Sahir]

Dear jorge

u alread assigned assigned to high availability cluster and mark volume group sharable from node1

vgchange -c cluster -S sharable vg_name

now no need to execute the same command from node2 also
u have to activate the volume group in shared mode thats all

vgchange -a s /dev/vgpro02data

thanks and regards

Sajjad Sahir

[jorge palma]

Hi.

The command:

vgchange -a s /dev/vgpro02indx

for each volume group, works fine. And i have already the cluster volumes actives and shareds.

When i try to run Serviceguard i get:

agnesi/#cmrunpkg SG-RAC
Unable to run package SG-RAC on node godel, the node switching is disabled
Unable to run package SG-RAC on node agnesi, the node switching is disabled
cmrunpkg: Unable to start some package or package instances

¿What have i to change on cluster definition to run the cluster package?

In addition: in a enviroment like this (Serviceguard + Oracle RAC), there is a way to increase a LV size (LV, after all, are the oracle datafiles on this env.) or create new LV inside the VG that the cluster manage?

When i try to resize LV or resize VG i get:

agnesi/#lvextend -L 165 /dev/vgpro02data/raw_pro02_users
lvextend: The volume group "/dev/vgpro02data" is active in Shared Mode.
Cannot perform configuration change.


Thanks.

[nightwich]

Olá Jorge !! ..

I already see that you put tha VG in Shared Mode !! ..

Now your second problem!!

To run the pkg you must enable de switching beetween nodes to do that just run the command:

cmmodpkg -e -n "node_name" "pkg_name"

Man cmmodpkg for more information!!


If you have any doubt please reply !! ..

Regards

[jorge palma]

Yes, I did

cmmodpkg -e -n "node_name" "pkg_name"

and works fine


Anyway there is something wrong on my configuration. RAC it's up on just one node:

agnesi/#cmviewcl -v

CLUSTER STATUS
clspro up

NODE STATUS STATE
godel up running

Cluster_Lock_LVM:
VOLUME_GROUP PHYSICAL_VOLUME STATUS
/dev/vglock /dev/dsk/c9t0d1 up

Network_Parameters:
INTERFACE STATUS PATH NAME
PRIMARY up LinkAgg1 lan901
PRIMARY up LinkAgg2 lan902
STANDBY up 1/0/1/1/0/6/1 lan7

NODE STATUS STATE
agnesi up running

Cluster_Lock_LVM:
VOLUME_GROUP PHYSICAL_VOLUME STATUS
/dev/vglock /dev/dsk/c11t0d1 up

Network_Parameters:
INTERFACE STATUS PATH NAME
PRIMARY up LinkAgg1 lan901
PRIMARY up LinkAgg2 lan902
STANDBY up 1/0/1/1/0/6/1 lan9

MULTI_NODE_PACKAGES

PACKAGE STATUS STATE AUTO_RUN SYSTEM
SG-RAC up (1/2) changing enabled no

NODE_NAME STATUS STATE SWITCHING
godel up running enabled

Script_Parameters:
ITEM STATUS MAX_RESTARTS RESTARTS INTERCONNECT NAME
Service up 0 0 crsp-srv

NODE_NAME STATUS STATE SWITCHING
agnesi starting starting enabled

Script_Parameters:
ITEM STATUS MAX_RESTARTS RESTARTS INTERCONNECT NAME
Service unknown 0 0 crsp-srv

I can not do anything with tradicional cm-comands, because it always get back:

agnesi/#cmrunpkg SG-RAC
Cannot start SG-RAC while it is starting on node agnesi.

Oracle Clusterware it is generating a lot of file similar with:

agnesi/#more /tmp/crsctl.3640
OCR initialization failed accessing OCR device: PROC-26: Error while accessing the physical storage Operating System error [Permissi
on denied] [13]


Thanks.

[nightwich]

Ola Jose !! ..

I think you have something wrong in your configurations !!



In a RAC Cluster Environment you need to have 2 instances in Oracle !!


So you need to have 2 PKG in a two node cluster !!

In your environment with would be SG-RAC1 for node godel and instance one and SG-RAC2 for node agnesi and instance2


I Suggest you HALT the entirely cluster and alter your configurations !! ..

Regards !! ..

[jorge palma]

Are you sure i need 2 packages for each node and no one multi-node package?.

[nightwich]

Hi Jorge,


Yes I'am shure, in a RAC environment there are two instances of the oracle, you have to configure 1 package for instance one and other package for instance two without failover.

The packages share the same VG ( VG in shared mode).

In my environment the SG, is used only for activate/desactivate the VG and Lv's. The ClusterWare do the oracle's work.

You have to proceed the installation instruction for oracle RAC.