hi, list
In short, a mixed version of 2.0.x with 1.x.x does not work.
Here is why:
In 2.0.0 -2.0.8, heartbeat will send a request node message to all alive nodes
to get node
lists. After getting replies from all alive nodes heartbeat will call
comm_now_up() to claim all nodes are now active.
However, 1.x.x does not have request/reply node message build into it, i.e. it will ignore the request message.
Therefore 2.0.8 will not get a reply from 1.x.x heartbeat therefore never come to a stage that it thinks all nodes are alive.
This bug affects a mixes version of 2.0.x with 1.x.x. Mixed versions of 2.0.x
will not be affected.
The future version of 2.0.x will fix the problem.
For more info about this bug, please check
http://old.linux-foundation.org/developer_bugzilla/show_bug.cgi?id=1544
-Guochun
Guochun Shi wrote:
The log from Patrick Begou shows the communication is dead among the
two nodes after their initial communication.
I tried the combination (1.2.3 & 2.0.8) with a simple IP address v1
style resource. The communication is fine, one node can see the other
node without problem.
However the resource cannot start or fail over correctly. Two 2.0.8
nodes v1 style resource work correctly though. I opened a bug for this
(#1544).
-guochun
Alan Robertson wrote:
Looks like something is broken between 1.2.x versions and 2.0.x versions
:-(. This is the 2nd complaint about this in the last week.
Subject:
Re: [Linux-HA] Heartbeat compatibility question
From:
Patrick Begou <[EMAIL PROTECTED]>
Date:
Fri, 06 Apr 2007 09:22:54 +0200
To:
General Linux-HA mailing list
To:
General Linux-HA mailing list
I have keep a log file, even if I do not find any usefull information
for my (low) level of knowledge. It is attached, if it can help.
Dean is the 2.0.8 heartbeat node (FC6 X86_64) wich produce this log.
Ekman is the 1.2.3 heartbeat node (Sarge Amd64)
The only thing I notice are the messages:
info: flow control disabled due to different version heartbeat
info: Status update for node ekman: status active
I launch several hb_takeover commands (returning without doing
anything) and finaly "/etc/init.d/heartbeat stop" (heartbeat -k
freeze) but nothing appears in the logs (or I do not see them as I do
not know what should be logged).
Heartbeat was ended by a "killall heartbeat" at the end of the log file.
About extra problems, yes they have nothing to deal with heartbeat.
It was just to give some feedback to heartbeat users wich could reach
into this situation.
Now heartbeat 1.2 runs fine and my HA cluster is available with one
node in Sarge and one node in FC6. I will setup the second node and
update heartbeat to 2.08 in the next weeks.
Patrick
___
Linux-HA mailing list
Linux-HA@lists.linux-ha.org
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
___
Linux-HA mailing list
Linux-HA@lists.linux-ha.org
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
___
Linux-HA mailing list
Linux-HA@lists.linux-ha.org
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
