Bạn đang xem bản rút gọn của tài liệu. Xem và tải ngay bản đầy đủ của tài liệu tại đây (6.89 MB, 436 trang )
12.3.1 Network requirements
Because data is continually replicated between the two cluster nodes, carefully
consider your networking topology. Figure 12-5 shows a diagram of a typical
SDMC High Availability environment.
Note that the SDMC provides network path redundancy through two connections
to each Flexible Service Processor (FSP) of a managed system. When the front
and back IPs of the FSP are connected through different networks to the SDMC
and one of the networks goes down, then the connection is automatically
established to the FSP from the same SDMC through the other network,
providing continuous availability to the user.
Connected to 9.0.0.30
Active Node
Passive Node
IBM Systems Director
Management Console
IBM Systems Director
Management Console
eth09.0.0.11
eth09.0.0.21
eth110.0.0.11
Floating IP
eth0:1- 9.0.0.30
br0bridge
br1bridge
eth09.0.0.10
eth110.0.0.10
eth110.0.0.21
Virtual
machines
Hypervisor
br0bridge
br1bridge
eth09.0.0.20
eth110.0.0.20
Hardware
Public LAN
Private LAN
Figure 12-5 High Availability network topology
308
IBM Systems Director Management Console: Introduction and Overview
Replication IP address
Because the replication IP address on each of the two SDMC nodes in the High
Availability cluster is used to constantly synchronize data, it is important that this
network connection be fast as possible.
Guidelines for the replication IP address
The guidelines for the replication IP address are:
If possible, use a separate private network for replication between the nodes
for better performance and place the two SDMCs within close range.
A high-speed Ethernet link, such as a crossover cable using gigabit Ethernet,
can be used to reduce latency for the initial and ongoing synchronization
between the active and passive SDMCs.
Network status IP address
Although having a network status IP address is optional, it is a best practice. The
nodes use this IP address to determine which node is connected to the network.
If they cannot contact each other, the node that can connect this IP address
becomes the active node.
If you do not specify a network status IP address, both nodes might become
active at the same time. If one node cannot connect to the other, it assumes that
the other node has lost network connectivity and makes itself the active node. If
both nodes have network connectivity but cannot connect to each other, both
nodes might become active.
If both nodes become active, changes might be made to both nodes, but all
changes made to one node are lost. When connectivity is restored, one node
becomes passive and the data on the active node is copied to it. Changes that
were made to the node that becomes the passive node are lost.
Guidelines for the network status IP address
The guidelines for the network status IP address are:
This address works best in domains where all nodes are in the same subnet.
Having the nodes in different subnets makes it more likely that both nodes
can ping the IP address, but they cannot communicate with each other.
Do not install a firewall rule that blocks ICMP traffic between the cluster nodes
and the network status IP instance.
Choose an address that the nodes can connect to only when they can
connect to the whole network.
Choose an address that is available most of the time.
Chapter 12. High availability and redundancy
309
Choose an address that each node can reach by a single path. If there are
multiple paths, it is possible that one path is down, causing the ping to fail,
when the node is online.
Do not use an address that either node uses, including the floating IP
address.
The gateway address for the SDMC is a good candidate to use as a network
status IP address.
Floating IP address and agent manager IP address
The floating IP address is always initially assigned to the active node and is
reassigned to the secondary node that takes over as the active node during a
failure. This is the address that users should always access so that they always
connect to the currently active node.
If you use Common Agent Services (CAS) agents, they use the floating IP
address specified as the agent manager IP address to communicate with the
nodes. For information about common agents, see the Common Agent topic in
the Systems Director Information Center at:
http://publib.boulder.ibm.com/infocenter/director/v6r2x/index.jsp
The agent manager uses the agent manager IP address to identify itself to the
CAS agents that it manages. If you only specify one floating IP address, it is used
as the agent manager IP address. If you specify more than one floating IP
addresses, you can specify which one is the agent manager IP address.
Guidelines for the floating IP address
The guidelines for the floating IP address are:
It must not be in use.
All systems that need to connect to the nodes must be able to access this IP
address through your network.
If your nodes are connected to multiple networks, configure multiple floating
IP addresses.
All CAS agents you want to manage must be able to contact the floating IP
address that you specify as the Agent Manager address.
310
IBM Systems Director Management Console: Introduction and Overview
12.3.2 DHCP considerations
The SDMC cannot be a DHCP client. If your network is set up so that one DHCP
server is running at a time (you have one range of DHCP addresses), then both
the SDMC network configurations must be identical and each network interface
card (NIC) must be connected to the same networks on each node. For example,
if eth0 is connected to the 192.168.1.0 network segment on the primary node,
then eth0 on the secondary node must also be connected to the same segment.
The SDMC High Availability environment can be configured in either a
non-shared or shared DHCP configuration, as shown in Figure 12-6 on
page 312.
Non-shared DHCP configuration
The DHCP server is run (and monitored) on each node. Configure each node to
use the same network interface (eth0 in Figure 12-6 on page 312) but to use
different address ranges.
Chapter 12. High availability and redundancy
311
Shared DHCP configuration
The DHCP server is only run on the active node. Configure the first node as
using DHCP server with a specific address range. Choose the first reserved
address for the node. Configure the second node as a standard network
configuration and choose the second reserved address for the node.
Non-shared DHCP Configuration
eth1
eth1
Shared DHCP Configuration
eth2
eth2
eth3
eth0
eth0
eth0
eth1
eth1
eth0
Switch
Switch
eth0
FSP NIC
eth3
eth1
eth1
eth0
eth0
FSP NIC
Switch
eth0
eth1
eth1
Switch
Figure 12-6 Shared and non-shared DHCP configurations
12.4 High Availability configuration
On the SDMC that is intended to be the primary node, perform the procedure in
this section to configure a high availability cluster in your environment. You must
have the SMAdministrator role to configure high availability.
312
IBM Systems Director Management Console: Introduction and Overview
12.4.1 Configuration using SDMC High Availability setup wizard
Perform the following steps:
1. Open up the IBM Systems Director Management Console GUI using a web
browser and log in as the sysadmin user. In the left navigation pane, expand
Settings and click High Availability Settings. The High Availability Settings
page opens. Click Set Up High Availability. Figure 12-7 shows the initial
page as you start the High Availability setup process.
Figure 12-7 High Availability Settings page
Note: If you do not see the Set Up High Availability button, high availability
is already configured.
Chapter 12. High availability and redundancy
313
2. The Welcome page for the High Availability setup wizard (Figure 12-8) shows
the SDMC as the primary node that you are using to perform the
configuration. It reminds the user of prerequisites and provides links to
support documents for more information about High Availability. Click Next to
begin setup.
Figure 12-8 Welcome page
314
IBM Systems Director Management Console: Introduction and Overview
3. In the Secondary Node setup page (Figure 12-9), enter the IP address or host
name and the password for the sysadmin user on that SDMC. Click Next.
Figure 12-9 Secondary Node page
4. In the Network Time Server page (not shown), specify the IP address for a
network time server or choose a primary node as the time server for both
nodes. Click Next.
Chapter 12. High availability and redundancy
315
5. In the Replication IP Address page (Figure 12-10), specify the IP addresses
to use for data replication between the two nodes (one on each node). Click
Next.
Figure 12-10 Replication IP Address settings page
316
IBM Systems Director Management Console: Introduction and Overview
6. If you want a network status IP address, determine an appropriate IP address
and enter it into the Network Status IP address page (Figure 12-11). Click
Next.
Note: A good choice for this address is the SDMCs gateway IP address.
Figure 12-11 Network Status IP address
Chapter 12. High availability and redundancy
317
7. Determine the available IP address to be used as the floating IP address and
enter this address at the Floating IP Address page (Figure 12-12). Click Next.
Figure 12-12 Adding floating IP addresses
318
IBM Systems Director Management Console: Introduction and Overview