AIX6.1 与HACMP6.1 ORACLE 10G RAC 系统上，再添加APPLICATION
已经安装完成power750 与ORACLE 10G . 再在用户要在上面添加一个APP应用，要求能进行切换，并绑一个IP地址。 思考了一下，发现有问题。 A、oracle 10g 采用crｓ方式进行配置。与IP地...显示全部
已经安装完成power750 与ORACLE 10G .
再在用户要在上面添加一个APP应用，要求能进行切换，并绑一个IP地址。
思考了一下，发现有问题。
A、oracle 10g 采用crｓ方式进行配置。与IP地址切换无关。只与节点上的IP地址切换有关。
B、再添加应用上去，需要配置IP地址切换。oracle 会用vip 冲突。
特别是本地网上切换时，oracle vip会报错，认为本地网卡出现问题。报错。
Configuring the IBM AIX 5L Operating System for the Oracle 10g VIP [ID 296856.1]
--------------------------------------------------------------------------------
修改时间 20-OCT-2010 类型 BULLETIN 状态 PUBLISHED
PURPOSE
-------
In order to avoid the public LAN from being a single point of failure, Oracle highly recommends
configuring a redundant set of public network interface cards (NIC's) on each cluster node. On
IBM AIX 5L platforms, it is our recommendation to take advantage of interface teaming technologies
provided by the AIX 5L operating system. By taking advantage of interface teaming technologies,
the AIX 5L operating system will handle the interface failure transparently from Oracle, thus the
VIP will continue to remain available to clients during interface failures.
As inter-node IP address failover is achieved by using the Oracle managed VIP, AIX / HACMP
based inter-node IP address failover technologies should not be configured on the same set of
NIC's that are used by the Oracle VIP. Only intra-node IP address failover functionalities
should be used in conjunction with the Oracle VIP.
This note will go over the basic configuration steps required to configure interface redundancy
on AIX 5L, plus the steps to configure the Oracle 10g VIP over of the redundant set of NIC's.
SCOPE & APPLICATION
-------------------
This article is intended for experienced DBAs and Support Engineers.
CONFIGURING INTERFACE TEAMING FOR THE VIP
-----------------------------------------
The Oracle 10g VIP may be configured on a redundant set of NIC's teamed together using NIC
teaming technologies. On AIX 5L, NIC teaming can be achieved by using either Network Interface
Backup (NIB) or EtherChannel. Both provide equivalent teaming functionalities, and are both
distributed as part of the AIX 5L operating system. While NIB is a feature that provides
minimum NIC failover capacities without requiring special hardware, EtherChannel is capable of
a richer variety of functionalities, such as link aggregation for higher bandwidth.
A typical EtherChannel configuration requires special network switches that are IEEE 802.3ad
compatible.
Below is an example of a typical NIB/EtherChannel configuration with two interfaces.
By using NIB/EtherChannel, Public LAN redundancy can be achieved at no additional software
cost, and single NIC failures will be handled transparently from Oracle within seconds,
meaning that the VIP will remain available and accessible with very little downtime during
single NIC failures.
NIB/EtherChannel
+----------------+
| Node A |
+--+----------+--+
en0 en1
| |
--- en0 ---
|
|
Please consult IBM for details on how to configure your system for NIC teaming.
Having configured NIC teaming correctly, a single NIC will appear in ifconfig, representing the
redundant set of NIC's teamed together. In the event of an underlying NIC failure, the
operating system will automatically route network requests to an available interface,
transparently from Oracle. Upon a total failure of all NIC's in the team, the VIP will no
longer be able to run on that node, and will fail over to a healthy cluster member. The NIC
name representing the redundant set of NIC's will remain unchanged during all events.
The NIC name representing the redundant set of NIC's (the teamed set of NIC's) is the NIC that
should be specified in the second screen in VIPCA (VIP Configuration Assistant, 1 of 2).
Make sure not to select any of the underlying non-redundant NIC names in VIPCA, as they should
not be used by Oracle in a NIC teaming configuration.
CONSIDERATIONS IN AN HACMP CONFIGURATION
----------------------------------------
HACMP IP address takeover is no longer required in Oracle 10g RAC to achieve high public LAN
availability, as the Oracle VIP handles inter-node IP address failover during failures.
If needed, HACMP Swap Adapter may be used in conjunction with the VIP. The following section
will go over the HACMP Swap Adapter configuration that can be used with the VIP.
CONFIGURING THE HACMP CLUSTERWARE FOR THE VIP
---------------------------------------------
When using HACMP clusterware, Swap Adapter technology may be used to achieve public LAN
redundancy. In this case, a set of NIC's will be configured in an active-standby configuration.
HACMP will manage a service IP address that will remain up and running on an available
interface. HACMP will relocate (or "swap") the IP address to the next available interface upon
an interface failure.
Below is an example of a typical HACMP Swap Adapter configuration with two interfaces.
Swap Adapter can be configured if the HACMP software is installed, and will maintain the
service IP on an available (primaty) interface (left diagram). Upon failure of the primary
interface, HACMP will swap the service IP to the standby interface.
HACMP Swap Adapter
+----------------+ +----------------+
| Node A | | Node A |
+--+----------+--+ +--+----------+--+
en0 en1 en0 en1
|(primary) |(standby) |(failed) |(primary)
| | ==========> | |
| | X |
| | | |
| | | |
| |
Service IP -- * Service IP -- *
When using HACMP Swap Adapter, the Oracle VIP will attempt to run on any available interface.
Note that Oracle does not make an attempt to stick to the "primary" interface managed by HACMP,
so there may be cases where the VIP is configured on the "standby" interface managed by HACMP.
The following are items that need to be carefully considered when configuring the VIP with
HACMP. Note that configuring Swap Adapter is not a requirement when using HACMP.
o Make sure that the service IP is only bound to a single node. This means that HACMP will
not fail over the service IP to a different cluster member upon network / node failures.
(The Oracle VIP will handle IP failover in 10g)
o In VIPCA, make sure to select all interfaces that constitute a Swap Adapter group where the
VIP should use. All "standby" interfaces need to be selected in addition to the active
interface, as the VIP may need to run on a "standby" interface upon an active interface failure.
o It is recommended to configure HACMP so that the lower-numbered NIC within the set of NIC's
is set to the initial "primary" interface. Though Oracle does not guarantee the VIP to always
run on the "primary" interface, making this configuration greatly reduces the chance of the VIP
coming up on the HACMP "standby" interface.
o Do not configure HACMP to relocate the hardware address (MAC address) in an adapter swap event.
When HACMP swaps an adapter, it will only relocate the service IP and the VIP address will be
unregistered. CRS will detect that the adapter has swapped at its next VIP check event, and will make
an attempt to bring up the VIP on an available NIC on that node. Note that in this event, the
VIP will become unavailable from clients until the next VIP check event. In order to avoid this
VIP outage, Oracle recommends using NIC teaming for higher availability.
(By default, CRS checks upon the VIP at every 60 seconds)
If a failure occurs on all NIC's, then the VIP will fail over to a healthy cluster member.
Some applications may monitor and restrict the outgoing / incoming interfaces for security
reasons. In such circumstances, applications may not be able to communicate correctly with
Oracle as the HACMP Swap Adapter configuration does not guarantee the VIP to run on the HACMP
"primary" interface.
(If the VIP happens to come up on the "standby" interface, then traffic may come in through the
"standby" and go out through the "primary", which may not be allowed by certain applications)
In this case, please use NIC teaming to achieve redundancy, instead of HACMP Swap Adapter.
RELATED DOCUMENTS
-----------------
For details on the networking technologies mentioned here, please refer to the following documentation
available on www.ibm.com:
o AIX 5L Version 5.2 System Management Guide: Communications and Networks >>
EtherChannel and IEEE 802.3ad Link Aggregation
o HACMP V5.2 Concepts and Facilities Guide >> Ch.5 Ensuring Application Availability >>
Eliminating Single Points of Failure in an HACMP Cluster >>
Eliminating Communication Interfaces as a Single Point of Failure收起
再在用户要在上面添加一个APP应用，要求能进行切换，并绑一个IP地址。
思考了一下，发现有问题。
A、oracle 10g 采用crｓ方式进行配置。与IP地址切换无关。只与节点上的IP地址切换有关。
B、再添加应用上去，需要配置IP地址切换。oracle 会用vip 冲突。
特别是本地网上切换时，oracle vip会报错，认为本地网卡出现问题。报错。
Configuring the IBM AIX 5L Operating System for the Oracle 10g VIP [ID 296856.1]
--------------------------------------------------------------------------------
修改时间 20-OCT-2010 类型 BULLETIN 状态 PUBLISHED
PURPOSE
-------
In order to avoid the public LAN from being a single point of failure, Oracle highly recommends
configuring a redundant set of public network interface cards (NIC's) on each cluster node. On
IBM AIX 5L platforms, it is our recommendation to take advantage of interface teaming technologies
provided by the AIX 5L operating system. By taking advantage of interface teaming technologies,
the AIX 5L operating system will handle the interface failure transparently from Oracle, thus the
VIP will continue to remain available to clients during interface failures.
As inter-node IP address failover is achieved by using the Oracle managed VIP, AIX / HACMP
based inter-node IP address failover technologies should not be configured on the same set of
NIC's that are used by the Oracle VIP. Only intra-node IP address failover functionalities
should be used in conjunction with the Oracle VIP.
This note will go over the basic configuration steps required to configure interface redundancy
on AIX 5L, plus the steps to configure the Oracle 10g VIP over of the redundant set of NIC's.
SCOPE & APPLICATION
-------------------
This article is intended for experienced DBAs and Support Engineers.
CONFIGURING INTERFACE TEAMING FOR THE VIP
-----------------------------------------
The Oracle 10g VIP may be configured on a redundant set of NIC's teamed together using NIC
teaming technologies. On AIX 5L, NIC teaming can be achieved by using either Network Interface
Backup (NIB) or EtherChannel. Both provide equivalent teaming functionalities, and are both
distributed as part of the AIX 5L operating system. While NIB is a feature that provides
minimum NIC failover capacities without requiring special hardware, EtherChannel is capable of
a richer variety of functionalities, such as link aggregation for higher bandwidth.
A typical EtherChannel configuration requires special network switches that are IEEE 802.3ad
compatible.
Below is an example of a typical NIB/EtherChannel configuration with two interfaces.
By using NIB/EtherChannel, Public LAN redundancy can be achieved at no additional software
cost, and single NIC failures will be handled transparently from Oracle within seconds,
meaning that the VIP will remain available and accessible with very little downtime during
single NIC failures.
NIB/EtherChannel
+----------------+
| Node A |
+--+----------+--+
en0 en1
| |
--- en0 ---
|
|
Please consult IBM for details on how to configure your system for NIC teaming.
Having configured NIC teaming correctly, a single NIC will appear in ifconfig, representing the
redundant set of NIC's teamed together. In the event of an underlying NIC failure, the
operating system will automatically route network requests to an available interface,
transparently from Oracle. Upon a total failure of all NIC's in the team, the VIP will no
longer be able to run on that node, and will fail over to a healthy cluster member. The NIC
name representing the redundant set of NIC's will remain unchanged during all events.
The NIC name representing the redundant set of NIC's (the teamed set of NIC's) is the NIC that
should be specified in the second screen in VIPCA (VIP Configuration Assistant, 1 of 2).
Make sure not to select any of the underlying non-redundant NIC names in VIPCA, as they should
not be used by Oracle in a NIC teaming configuration.
CONSIDERATIONS IN AN HACMP CONFIGURATION
----------------------------------------
HACMP IP address takeover is no longer required in Oracle 10g RAC to achieve high public LAN
availability, as the Oracle VIP handles inter-node IP address failover during failures.
If needed, HACMP Swap Adapter may be used in conjunction with the VIP. The following section
will go over the HACMP Swap Adapter configuration that can be used with the VIP.
CONFIGURING THE HACMP CLUSTERWARE FOR THE VIP
---------------------------------------------
When using HACMP clusterware, Swap Adapter technology may be used to achieve public LAN
redundancy. In this case, a set of NIC's will be configured in an active-standby configuration.
HACMP will manage a service IP address that will remain up and running on an available
interface. HACMP will relocate (or "swap") the IP address to the next available interface upon
an interface failure.
Below is an example of a typical HACMP Swap Adapter configuration with two interfaces.
Swap Adapter can be configured if the HACMP software is installed, and will maintain the
service IP on an available (primaty) interface (left diagram). Upon failure of the primary
interface, HACMP will swap the service IP to the standby interface.
HACMP Swap Adapter
+----------------+ +----------------+
| Node A | | Node A |
+--+----------+--+ +--+----------+--+
en0 en1 en0 en1
|(primary) |(standby) |(failed) |(primary)
| | ==========> | |
| | X |
| | | |
| | | |
| |
Service IP -- * Service IP -- *
When using HACMP Swap Adapter, the Oracle VIP will attempt to run on any available interface.
Note that Oracle does not make an attempt to stick to the "primary" interface managed by HACMP,
so there may be cases where the VIP is configured on the "standby" interface managed by HACMP.
The following are items that need to be carefully considered when configuring the VIP with
HACMP. Note that configuring Swap Adapter is not a requirement when using HACMP.
o Make sure that the service IP is only bound to a single node. This means that HACMP will
not fail over the service IP to a different cluster member upon network / node failures.
(The Oracle VIP will handle IP failover in 10g)
o In VIPCA, make sure to select all interfaces that constitute a Swap Adapter group where the
VIP should use. All "standby" interfaces need to be selected in addition to the active
interface, as the VIP may need to run on a "standby" interface upon an active interface failure.
o It is recommended to configure HACMP so that the lower-numbered NIC within the set of NIC's
is set to the initial "primary" interface. Though Oracle does not guarantee the VIP to always
run on the "primary" interface, making this configuration greatly reduces the chance of the VIP
coming up on the HACMP "standby" interface.
o Do not configure HACMP to relocate the hardware address (MAC address) in an adapter swap event.
When HACMP swaps an adapter, it will only relocate the service IP and the VIP address will be
unregistered. CRS will detect that the adapter has swapped at its next VIP check event, and will make
an attempt to bring up the VIP on an available NIC on that node. Note that in this event, the
VIP will become unavailable from clients until the next VIP check event. In order to avoid this
VIP outage, Oracle recommends using NIC teaming for higher availability.
(By default, CRS checks upon the VIP at every 60 seconds)
If a failure occurs on all NIC's, then the VIP will fail over to a healthy cluster member.
Some applications may monitor and restrict the outgoing / incoming interfaces for security
reasons. In such circumstances, applications may not be able to communicate correctly with
Oracle as the HACMP Swap Adapter configuration does not guarantee the VIP to run on the HACMP
"primary" interface.
(If the VIP happens to come up on the "standby" interface, then traffic may come in through the
"standby" and go out through the "primary", which may not be allowed by certain applications)
In this case, please use NIC teaming to achieve redundancy, instead of HACMP Swap Adapter.
RELATED DOCUMENTS
-----------------
For details on the networking technologies mentioned here, please refer to the following documentation
available on www.ibm.com:
o AIX 5L Version 5.2 System Management Guide: Communications and Networks >>
EtherChannel and IEEE 802.3ad Link Aggregation
o HACMP V5.2 Concepts and Facilities Guide >> Ch.5 Ensuring Application Availability >>
Eliminating Single Points of Failure in an HACMP Cluster >>
Eliminating Communication Interfaces as a Single Point of Failure收起