Installing the HA Software
First, install the software
The most stable version is 5.4.0, priority installation stable version 5.4
Install depends on package base.data, cluster man pack installation failure reason is missing Base.data package
Install all cluster. Packages that begin with, you can not install the Cluster.man package
Second, hit the HA patch
Execute the instructions before patching: Inutoc generate the. toc file, and then Smitty Update_all upgrade Cluster to 5.4.1.0 version
Considerations for building HA Environments
First, the HA configuration is the primary standby mode or the interoperability mode should be based on the requirements of the application system to choose, when set the priority of the resource run, the primary node is the preferred running node of a resource group.
Second, node switch does not need VARYOFFVG, ha can automatically varyoffvg and mount the FS
Third, before switching oradatavg, you must first shutdown the Oracle, otherwise Oracle will be down
Four, when the different nodes switch VG, before the first switch to the new node must be executed IMPORTVG, the VG added to the Aix ODM library, so that HA can varyonvg, otherwise cannot find the path of the resource, if you want to back-cut resources, but previously executed on node EXPORTVG That is, modify the VG specified path in the ODM, this is not VARYONVG, you must re-execute the IMPORTVG instruction to add the path of the VG to the ODM library, ha to automatically activate the VG.
V. When the HA switches between node, the VG is only responsible for VARYONVG and VARYONFFVG not responsible for IMPORTVG and EXPORTVG
Vi. free switching between HA nodes does not require OS password authentication and is completed by the system daemon
Vii. before deleting cluster, you must first stop the Cluster service for all nodes, or the cluster will crash
Viii. when synchronizing information between cluster nodes, the contents of the cluster configuration file are synchronized, not the contents of the OS on each node.
Build Oracle Ha Steps
First, install Oracle software
1.
Ii. DBCA Create Oracle Library
III. Configuring cluster resources (service IP)
Iv. Configuring cluster resources (ORADATAVG)
V. Configuring cluster resources (APP server)
Vi. switching between different node Oracle
Solving the problem of brain fissure in ha
First, install the package that is required to create the concurrent VG CLVM package, the package installation, upgrade, after the OS must be restarted
Description of the CLVM package: Enhanced Concurrent Logical Volume Manager
Package in Aix6100-dvd1.iso: Install to go to the INSTALLP/PPC directory to perform the installation
Package upgrade in 6106: Upgrade using the directive Smitty Update_all directly Select all to upgrade to the latest version, do not support the selection of some package upgrades, the system only supports all related packages upgrade
Second, identify shared storage
There are three ways to determine shared storage:
Method One:
On a shared storage server LSPV, if the corresponding HDISKN Pvid is the same, it can be determined to be a shared storage
Additional execution instructions: Bootinfo-s hdiskn, to see the size of the corresponding storage, the size of the shared storage must be the same
Method Two:
Only for IBM storage to take effect, only for IBM DS system Storage
Execution instruction: Mpio_get_config-av
The value of the User label column in the display result is unique, and the instruction is executed concurrently on the 211 and 212 machines.
If the User lablel column value of the corresponding HDISKN is the same, the shared storage
Method Three:
Use the Dbh_read directive to make sure that no heartbeat disk is on the shared storage before using this instruction, otherwise the test information is written to the disk header of the shared storage
On 211 End:/usr/sbin/rsct/bin/dhb_read-p Hdisk8-r Enter information receive status
On 212 end:/usr/sbin/rsct/bin/dhb_read-p hdisk8-t Enter information transate status
If the test information sent on the 211 end can be successfully received, it means that the HDISK8 disk is a shared storage
Third, create a concurrent VG, do disk heartbeat
1, under 211 to create HBORAVG type enhanced concurrent added concurrent VG, while choosing not automatic active, concurrent VG do disk heartbeat just use the Vgda area of the concurrent VG, use only the disk header, not Set the system to automatically activate when rebooting,
2. Under 211: Create test LV under concurrent VG to test whether the concurrent VG was created successfully, that is, whether the VG can be used for concurrent VG, LV name: SYCHLV
You need to ACTIVEVG:VARYONVG Hboravg before creating a LV
Then create the Lv:smitty mklv
3, 211 under:
Deactive VG:VARYOFFVG Hboravg
4. Import at 212:
Import Vg:importvg-v mainnum-y Hboravg
The LV in the VG can be displayed normally at 212, usually the type of LV is not displayed?? It can be judged as normal.
5, 211 and 212 end all deactive VG
Execution instruction: VARYOFFVG hboravg
Four, start the cluster, activate the GSCLVMD process, through the process to the active concurrent VG, test the VG can be synchronized
1, start the cluster, because the VARYONVG-C HBORAVG activation to enhanced concurrent mode must have a GSCLVMD process, but the GSCLVMD process is initiated by HA, so before activation must have a GSCLVMD process exists
View process directive: PS-EF | grep GSCLVMD
2. Activate VG to enhanced concurrent mode
211 and 212 End all execution instructions: varyonvg-c hboravg
The status displayed after the VG is activated is not active but concurrent
3, test whether the concurrent VG
Create the TESTLV on the 211 end, the normal situation should be automatically synchronized to 212 end
The condition of the VG is shown on the 212 end: Lsvg-l hboravg
Should be able to display the LV created on the 211 end, which is automatically pushed to the 212 end
Under 212:
Cd/dev
Ls-l *test*
device files corresponding to VG and LV can be displayed normally
V. Add concurrent VG to ha to test if the heartbeat is successful
1, directly add configure HACMP communication interfaces/devices
Choice: Devices
Select: aix211 hdisk8 and aix212 hdisk8
2. Synchronous cluster
3. Testing
(i) the resource group is currently running on 211,
Ifconfig en0 down: Resource cut to 211 on the EN1 NIC
Ifconfig en1 down: Resources should switch to 212 En0, while Seviceip disappears on 211
(ii) Under 212, review and verify that the resource switch was successful and that the heartbeat was successful
On 212 end: Ifconfig-a |service IP resources should be able to successfully switch to 212
Restart 211 of the two network cards En0 and EN1, the resources are still running on 212 nodes, and 212 nodes will not panic, should be able to take over the resource group.
Vi. solving the problem of ha brain crack
Summarize:
The sequence of the entire process operation:
Create HBORAVG----> Create cluster--->add node--->descover---->add network-->add interface--
-->config Resource--->add resource Group---> Resources loading show/change----> synchronizing clusters to each node----> starting the cluster---- Varyonvg-c HBORAVG boot to enhanced concurrent mode---
----> Create Test LV Verify concurrent VG----Add Device HDISK8 (heartbeat disk)----> Sync cluster to each node
--->down network card to test if the heart drive is active
This article is started by the shrimp "Diffuse network (http://www.manxinet.com)" without permission to be reproduced in any way, offenders will be held liable
Ha high-availability cluster cluster under AIX