Migrate configurations from Load Balancer IPv4 to Load
Balancer IPv4 and IPv6.
About this task
Migrate configurations from Load Balancer IPv4 to Load Balancer
IPv4 and IPv6, as follows:
Procedure
- Search for method CBR protocol, "<http | ssl>",
"cookiename", "cookievalue", "content", "pattern", "affinity uri"
->. If any exist, CBR forwarding is being used,
making migration difficult. Consider using plug-in, VE, or CP+CBR.
This type of forwarding is not supported by Load Balancer for IPv4
and IPv6.
- Search for wideportnumber. If
found, forwarding to another load balancer is not supported. Forward
directly to backend servers. Migration might be complicated because
you must obtain the configurations from two load balancers and merge
them.
- Search for subagent. If
present, remove and install and configure an external SNMP monitor
agent.
- Search for maxhalfopen <value>
or halfopenaddressreport. If present,
remove and perform Denial of Service at the firewall and possibly,
backend servers.
- Look for mapport. If present,
configure port translation at the operating system. If this option
is not possible, consider using CP/CBR, plug-in, or VE.
- Change all occurrences : to @ in
the configuration.
- Remove any executor configure or executor
unconfigure lines (remove the entire line). Check go
scripts and remove any aliasing or unaliasing. The Go
scripts command logs only the takeover events, if used.
- Search for method nat on port
add commands. If present, remove method
nat and add encapforward yes encaptype <ipip
| gre > encapcond always or nat yesto all
of the server adds for this port, where the server is on a remote
subnet. You might have to know the host name to IP resolution to know
whether the hosts are located remotely. If you do not have any reverse
path filtering enabled that would block the request, the encapforward
yes command performs better and is recommended over NAT.
If you are using the encapsulated forwarding, configure the backend
machines to decapsulate. You cannot use encapsulated forwarding to
backend Windows computers.
- Search for highavailability backup add both. MHA is not supported. Determine which load balancer is primary.
Convert the commands to indicate primary and backup on respective
configurations. Determine repstrategy and set
it on the portcommands.
- Determine how you set server selection algorithms on the port commands.
- Search for: primaryhost <ip>, weightbound <number>,
maxclusters <number>, maxport <number>, maxservers <number>,
porttype <udp |tcp | both>, sharedbandwidth <value>, fintimeout <value>,
mss <value>, hasynctime <value>. If found,
inspect the command. These commands might be the only specification
on a set command and if so, delete the entire
line.
- Search for stickytimeor staletimeout. If present on an executor or cluster command, move it to the
ports that are added after that executor or cluster command that does
not already have a stickytime or staletimeout on
the add or set command.
- If the configuration file shows cbrcontrol,
instead of dscontrol, you are using the CBR method
not dispatcher. Change all instances of : to @.
Use the CBR component for IPv6 instead of the CBR component for IPv4.
The same method applies if the configuration shows sscontrol.
Use Site Selector from IPv6.
CAUTION:
If the
configuration shows ccocontrol or nalcontrol,
the hardware is not supported.
Note: The ip, time, port
and service rule types are not supported.