This example shows the steps to allow the DMP to use as multipathing software for VCS LVMVG SG on AIX.
Step 1: Ensure Enclosure Based Name (EBN) name is used and dmp_native_support is ON
vxddladm get namingscheme
vxddladm set namingscheme=ebn [persistence={yes|no}] [use_avid=yes|no] [lowercase=yes|no]
vxdmpadm gettune |grep dmp_native_support
vxdmpadm settune dmp_native_support=on
Step 2: Remove device from from VxVm and create LVM VG on node1
vxdisk rm emc0_25
chpv -C emc0_25 (clear PVID if any)
mkvg -f -y vmaxlvmvg emc0_25
mklv -t jfs2log vmaxlvmvg 1
mkfs -o log=/dev/loglv01 -V jfs2 /dev/vmax_lvm
mklv -t jfs2 -y vmaxlvmlv vmaxlvmvg 1G
crfs -v jfs -m <mount point> -d <logical volume>
lspv |grep vmaxlvmvg (identify PVID)
Step3: set physical volume (pv) to the other node2
chdev -l emc1_25 -a pv=yes
lsattr -El emc1_25
Step 4: Export/Import VG on both cluster node
exportvg vmaxlvmvg (node1)
lvlstmajor
varyoffvg vmaxlvmvg
importvg -V major_number -y vmaxlvmvg emc0_25
varyoffvg vmaxlvmvg
lvlstmajor (node 2)
importvg -V major_number -y vmaxlvmvg emc1_25
varyoffvg vmaxlvmvg
If there's any problem bring online under VCS, refer to http://www.symantec.com/business/support/index?page=content&id=TECH70949 for troubeshooting.
Note: If third party driver (TDPmode) eg. PowerPath is used for multipathing. Step 1 can be skipped and PP pseudo name is normally used for VG creation, ie mkvg -y vmaxlvmvg hdiskpower25. Refer to http://sfdoccentral.symantec.com/sf/5.1SP1PR1/aix/pdf/dmp_admin_51sp1pr1_aix.pdf page 154 for more details.
This blog is created to keep track of issues encountered for clustering software such as VCS, SFRAC, Oracle RAC, HP MetroCluster and LifeKeeper. Hopefully, this blog will help whoever encounters the similar problem. This site also provides relevant links which might be useful.
Monday, September 5, 2011
VCS SG Linux LVM failure due to all paths disabled
There is a limitation on VCS LVMVolumeGroup agent and Linux LVM causing SG LVM SG cannot failover to other node when there's all path disabled on the active node. It is a known issue which is documented.
Refer to VCS release notes https://sort.symantec.com/public/documents/sfha/5.1sp1pr2/linux/productguides/pdf/vcs_notes_51sp1pr2_lin.pdf page 43 for more details.
Refer to VCS release notes https://sort.symantec.com/public/documents/sfha/5.1sp1pr2/linux/productguides/pdf/vcs_notes_51sp1pr2_lin.pdf page 43 for more details.
Thursday, September 1, 2011
Oracle RAC 11gR2 Issues
This post is some of the oracle articles I found in the Oracle support knowledge base website during my Oracle RAC 11gR2 installation. You might find them useful if you encounter similar issue. Take note a login account is required to access the website.
Article ID
CLUVFY Fails with TCP Check PRVF-7617 Due to Case of Node Names [ID 1286394.1]
Troubleshooting 11.2 Grid Infastructure Installation Root.sh Issues [ID 1053970.1]
CRS Does not Start after Node Reboot in 11gR2 Grid Infrastructure [ID 1215893.1]
Grid Infrastructure Installation: root.sh Stalls Due to Improper localhost Setting [ID 1155903.1]
Recommendation for the Real Application Cluster Interconnect and Jumbo Frames [ID 341788.1]
Article ID
CLUVFY Fails with TCP Check PRVF-7617 Due to Case of Node Names [ID 1286394.1]
Troubleshooting 11.2 Grid Infastructure Installation Root.sh Issues [ID 1053970.1]
CRS Does not Start after Node Reboot in 11gR2 Grid Infrastructure [ID 1215893.1]
Grid Infrastructure Installation: root.sh Stalls Due to Improper localhost Setting [ID 1155903.1]
Recommendation for the Real Application Cluster Interconnect and Jumbo Frames [ID 341788.1]
Oracle Cluster Verification Utility
Oracle provides Cluster node verification prior to installation. It is very helpful for Oracle RAC pre-installation to check the nodes meeting the system minimum requirement, the network interfaces availability, the patches update and etc.
Oracle Grid pre-installation
To verify whether your system meets all of the criteria for an Oracle Clusterware installation:
Window:
After you have completed phase one, verify that Oracle Clusterware is functioning properly before proceeding with phase two of your Oracle RAC installation:
Window:
To verify whether your system meets all of the criteria for an Oracle RAC installation:
Window:
Oracle Grid pre-installation
To verify whether your system meets all of the criteria for an Oracle Clusterware installation:
Window:
cluvfy stage -pre crsinst -n node1,node2,node3
Unix:
runcluvfy.sh stage -pre crsinst -n node1,node2,node3
Window:
cluvfy stage -post crsinst -n node1,node2,node3 -vUnix:
runcluvfy.sh stage -post crsinst -n node1,node2,node3 -vWindow:
cluvfy stage -pre dbinst -n node1,node2,node3 -v Unix:
Oracle Database pre-installation
To verify whether your system meets all of the criteria for creating a database or for making a database configuration change:
Window:cluvfy stage -pre dbcfg -n -n node1,node2,node3 -d oracle_home -v
Unix:
runcluvfy.sh stage -pre dbcfg -n -n node1,node2,node3 -d oracle_home -v
Refer to Cluster Verification Utility Reference for more detail explanation.
runcluvfy.sh stage -pre dbinst -n node1,node2,node3 -v
Veritas DMP HP-UX native LVM configuration
HP-UX native LVM is supported with Veritas DMP for multipathing. dmp_native_support parameter must be turned on and Enclosure Base Name (EBN) scheming must be used in order for DMP to manage LVM.
#vxddladm get namingscheme
#vxddladm set namingscheme=ebn persistence=yes
#vxdmpadm gettune
#vxdmpadm settune dmp_native_support=on
#vxddladm get namingscheme
#vxddladm set namingscheme=ebn persistence=yes
#vxdmpadm gettune
#vxdmpadm settune dmp_native_support=on
On HP-UX 11.31 1003IC, the pvcreate Utility fails to create Physical
volumes with DMP devices. As a work around, use corresponding native multipathing (nMP metanode/OS path) to create physical volume using pvcreate Utility.
Instead of "pvcreate -f /dev/vx/dmp/emc_clariion0_1", "pvcreate -f /dev/rdisk/disk110" should be used.
Subscribe to:
Posts (Atom)