Archide log (Oracle RAC) location using SLVM

[jorge palma]

Hi, I would like ask a question about a Oracle RAC installation (and conf.) using SGeRAC extension.

I have a RAC database using raw devices over SLVM Volume Groups. Archived logs are located on local disks
for each node. So, there is no share location for archived logs.

Yesterday i was trying to do


1) create a SLVM VolumeGroup (named vgarch01) shareable.
2) create a LV inside this VG. (named /dev/vgarch01/raw_pro02_arch01)
3) For each node i executed:

newfs -F vxfs -o largefiles -b 8192 /dev/vgarch01/rraw_pro02_arch01
mkdir /ARCH
mount /dev/vgarch01/raw_pro02_arch01 /ARCH/


But the nodes are not sharing the "location", cause:


(agnesi it's node 1)


agnesi/#hostname
agnesi
agnesi/#ls -l /ARCH/
total 0
-rw-rw-rw- 1 root sys 0 Aug 6 14:20 agnesi
drwxr-xr-x 2 root root 96 Aug 6 14:20 lost+found


(godel it's node 2)

godel/#hostname
godel
godel/#ls -l /ARCH
total 0
drwxrwxrwx 2 root sys 96 Aug 6 14:23 GODEL
drwxr-xr-x 2 root root 96 Aug 6 14:19 lost+found



Is there something wrong on my config? Could i locate in a share directory archived log destinations?.


ENVIROMENT:

agnesi/#uname -a
HP-UX agnesi B.11.31 U ia64 1234912541 unlimited-user license

SG Version: A.11.18.00

Thanks.

[Steven E. Protter]

Shalom,

For the same systems to archive log on the same shared storage at the same time, the volume group must be activated in shared mode.

vgchange -a s

The volume group needs to be created on one node.

vgexport -p -m must be used to create a map file.

The map file must be moved to the second node and vgimport must be used to create the node.

If the volume group is not activated in shared mode or the volume group has not been migrated via vgexport/vgimport then this is probable cause.

It might be less complex to have a volume group dedicated to reach system and not to share. That is your choice.

SEP

[jorge palma]

But Volume Group it's already on shared mode:

agnesi/#vgdisplay -v vgarch01
--- Volume groups ---
VG Name /dev/vgarch01
VG Write Access read/write
VG Status available, shared, server
Max LV 255
Cur LV 1
Open LV 1
Max PV 16
Cur PV 1
Act PV 1
Max PE per PV 1016
VGDA 2
PE Size (Mbytes) 16
Total PE 63
Alloc PE 63
Free PE 0
Total PVG 0
Total Spare PVs 0
Total Spare PVs in use 0
VG Version 1.0
VG Max Size 254g
VG Max Extents 16256

--- ---

agnesi Server
godel Client

--- Logical volumes ---
LV Name /dev/vgarch01/raw_pro02_arch01
LV Status available/syncd
LV Size (Mbytes) 1008
Current LE 63
Allocated PE 63
Used PV 1


--- Physical volumes ---
PV Name /dev/dsk/c9t1d1
PV Name /dev/dsk/c11t1d1 Alternate Link
PV Status available
Total PE 63
Free PE 0
Autoswitch On
Proactive Polling On

[Duncan Edmonstone · Accepted solution]

What made you think you can create a VxFS filesystem on a shared volume an access it concurrently from more than 1 node?

You can't - unless you are using the Cluster Filesystem, and for that you need the Serviceguard Storage Management Suite for Oracle RAC:

http://www.hp.com/go/sgsms

In a SLVM environment your choices are basically:

i) if you are on 10gR2 or above you could convert you database to running on ASM over SLVM - see this WP here:

http://docs.hp.com/en/8988/ASM-SGeRAC-tk.pdf

In an ASM configuration, the archive logs exist in the ASM storage on the raw disks rather than on a filesystem.

ii) If you want to stay with byour existing configuration, then you'll need to use HA NFS packages to make sure the archive log areas are highly available as NFS mounts across your cluster.

HTH

Duncan

[jorge palma]

Ok. That's what i was waiting to read.

So, How could i know if i have licensed Cluster Volume Manager (CVM)?

swlist o.s command could do it?.

Thanks in advance.

[Duncan Edmonstone]

It's not just the CVM (Cluster Volume Manager), but CFS as well (Cluster Filesystem).

I can't tell you if you are *licensed* for it or not (you'd need to look at your original quote/PO to figure that out), but I can tell you that if you have the "cfscluster" command on your system then you have it installed.

HTH

Duncan