I'm trying to migrate an existing two cluster r1 setup (A,B) to another two
cluster setup (C,D).  The catch is C and D both have different ip addresses
but want to have them end up with the same hostname as A/B.
Basically I want to move the vip that A,B use to the new set (C,D) and keep
the names.
I should point out that A/B and C/D each pair is hardwired to each other
via eth1.

A <-----> B
C <-----> D

Ips are A=10.0.0.10, B=10.0.0.20, C=10.0.0.30, D=10.0.0.40

I'd like to know how to do this without incurring any downtime if possible.
Here is the procedure I've been trying to use to migrate from A to C:

1. On A, run hb_standby all  (moves vip to B)
2. On A, heartbeat stop
3. On A, copy hb_generation, hb_uuid to C
4. On C, run 'hostname A' command
5. Update dns to change ip for A to C
6. On C, heartbeat start

I've tried this multiple times and it has worked sometimes but most of the
time, heartbeat on C starts but when running 'cl_status rscstatus' it says
it cannot connect.  When it has worked, running cl_status rscstatus on C
will show 'none' like i would expect it since the vip is still on B.

Note I put a hostname (instead of an ip) in the ucast
line for A. The behavior can vary as well as sometimes it may have
the vip on B and C after I start heartbeat on C.  Given R1 limitations
to two nodes only, I'm not sure how else to approach this.  I saw this in
the faq:

http://www.linux-ha.org/FAQ#head-c86eb68ed7d45019036040f6724d483ced91b483

Not exactly what I'm trying to do but I tried putting all 4 hosts as node
names in ha.cf but when I restarted heartbeat it didn't seem to like it.

A few questions:

- Is it ok to put a hostname in the ucast line?
        - If not, that means I will have to put the IP for A and will want to 
change
          it to the ip of C, for the migration from B to C.
        - If so, do I need to reload the config so it does another dns lookup 
to get
          the updated ip?

- Is there a better way to do this using an R1 config?

Here are my configs:

ha.cf (A):

ucast bond0 10.0.0.10
bcast eth1
logfacility daemon
keepalive 2
warntime 10
deadtime 30
deadping 30
initdead 120
udpport 694
auto_failback no
use_logd yes

node A
node B

respawn hacluster /usr/lib64/heartbeat/ccm
respawn hacluster /usr/lib64/heartbeat/ipfail

ping 10.0.0.1


ha.cf (B)

ucast bond0 A
bcast eth1
logfacility daemon
keepalive 2
warntime 10
deadtime 30
deadping 30
initdead 120
udpport 694
auto_failback no
use_logd yes
debug 3

node A
node B

respawn hacluster /usr/lib64/heartbeat/ccm
respawn hacluster /usr/lib64/heartbeat/ipfail

ping 10.0.0.1

haresources:

A sleep1 10.0.0.55

_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to