Discussion:
heartbeat over more than one NIC, mcast explanation
Gregor Goestl
2003-01-18 17:19:44 UTC
Permalink
I have two questions:

First:
Is it possible to use more than one NIC for heartbeat signals? Does the
following setting in ha.cf work:

udp eth0
udp eth1

to send heartbeats over two different ethernet connections?

Second:
Can me someone explain me the functionality of the 'mcast' Setting in ha.cf?
What can I do with that stuff?

Thanks!
Bye, Gregor
Brian Tinsley
2003-01-19 02:51:47 UTC
Permalink
Post by Gregor Goestl
Is it possible to use more than one NIC for heartbeat signals? Does the
udp eth0
udp eth1
to send heartbeats over two different ethernet connections?
Yes
Post by Gregor Goestl
Can me someone explain me the functionality of the 'mcast' Setting in ha.cf?
What can I do with that stuff?
the syntax in ha.cf is as follows:

mcast [dev] [mcast group] [port] [ttl] [loop]

[dev] device to send/rcv heartbeats on
[mcast group] multicast group to join (class D multicast address
224.0.0.0 - 239.255.255.255)
[port] udp port to sendto/rcvfrom (no real reason to differ
from the port used for broadcast heartbeats)
[ttl] the ttl value for outbound heartbeats. this effects
how far the multicast packet will propagate. (0-255)
[loop] toggles loopback for outbound multicast heartbeats.
if enabled, an outbound packet will be looped
back and
received by the interface it was sent on. (0 or 1)
Gregor Goestl
2003-01-19 07:43:52 UTC
Permalink
----- Original Message -----
From: "Brian Tinsley" <btinsley at emageon.com>
To: "Gregor Goestl" <gregor.goestl.da at mystic.cc>
Cc: "Linux-HA" <linux-ha at muc.de>
Sent: Sunday, January 19, 2003 3:51 AM
Subject: Re: heartbeat over more than one NIC, mcast explanation
Post by Brian Tinsley
Post by Gregor Goestl
Is it possible to use more than one NIC for heartbeat signals? Does the
udp eth0
udp eth1
to send heartbeats over two different ethernet connections?
Yes
Post by Gregor Goestl
Can me someone explain me the functionality of the 'mcast' Setting in ha.cf?
What can I do with that stuff?
mcast [dev] [mcast group] [port] [ttl] [loop]
[dev] device to send/rcv heartbeats on
[mcast group] multicast group to join (class D multicast address
224.0.0.0 - 239.255.255.255)
[port] udp port to sendto/rcvfrom (no real reason to differ
from the port used for broadcast heartbeats)
[ttl] the ttl value for outbound heartbeats. this effects
how far the multicast packet will propagate.
(0-255)
Post by Brian Tinsley
[loop] toggles loopback for outbound multicast heartbeats.
if enabled, an outbound packet will be looped
back and
received by the interface it was sent on. (0 or 1)
I had read this description in the ha.cf example, but I don't understand it.
What can I do with this setting?
Lars Marowsky-Bree
2003-01-19 16:34:30 UTC
Permalink
On 2003-01-19T08:43:52,
Post by Gregor Goestl
I had read this description in the ha.cf example, but I don't understand it.
What can I do with this setting?
Instead of using UDP broadcast for heartbeat's communication (like 'udp'
does), which is a) limitted to the local network segment and also can lead to
collisions more easily with multiple heartbeat clusters on the same wire, you
can make heartbeat communicate over a multicast group.

For example, http://www.google.com/search?q=Multicast+introduction would prove
to be a good starting point, yielding
http://www.online.kth.se/courses/common/multicast/intro1.html right as the
first answer ;-)


Sincerely,
Lars Marowsky-Br?e <lmb at suse.de>
--
Principal Squirrel
SuSE Labs - Research & Development, SuSE Linux AG

"If anything can go wrong, it will." "Chance favors the prepared (mind)."
-- Capt. Edward A. Murphy -- Louis Pasteur
Brian Tinsley
2003-01-19 16:56:03 UTC
Permalink
Post by Gregor Goestl
I had read this description in the ha.cf example, but I don't understand it.
What can I do with this setting?
And to add to what Lars posted, an example of usage in ha.cf might be:

mcast eth0 224.1.2.3 694 1 1

This would cause heartbeat to send and receive multicast UDP heartbeats
on the network device eth0. All nodes participating in this cluster will
also have to join the multicast group 224.1.2.3 (by also having this
line in their local ha.cf file); think of this as a "targeted
broadcast". All heartbeats will be sent and received on UDP port 694.
Setting the time-to-live to 1 will limit the multicast packets to the
local network segment (i.e., they will not propogate through routers to
other networks). Setting the loop parameter to 1 will cause a sent
heartbeat to also be received by the sender; I believe this is desirable.
Loading...