[Pacemaker] Nodes unable to connect / find each other

Thu Mar 15 14:24:51 UTC 2012

That is the weird part I do not understand.
Netstat shows the ports and process name, SELinux is disabled, local firewall is disabled, no other log entries other than those posted in initial mail.

Which is why I do not understand why it's not working.

UDP traffic only appears with multicast enabled and the bindnetaddress being the respective nodes' IPs. 
With unicast there's no traffic, nor is there traffic with multicast and bindnetaddress being 10.26.29.0

Cheers,

Achim

-----Original Message-----
From: Andreas Kurz [mailto:andreas at hastexo.com] 
Sent: 15 March 2012 13:27
To: pacemaker at oss.clusterlabs.org
Subject: Re: [Pacemaker] Nodes unable to connect / find each other

On 03/15/2012 01:57 PM, Regendoerp, Achim wrote:
> As a status update, not got any further...
> Confirmed with the Networks people that Multicast is enabled, but no luck.
> Using unicast crm is not able to connect to the cluster (below is the unicast config used).
> With unicast, there's no traffic at all (observed via tcpdump)

No traffic at all? And the network ports 5404/5405 are allocated when checking with netstat or lsof? selinux enabled and showing problems in audit logs, firwall issue?

Regards,
Andreas

--
Need help with Pacemaker?
http://www.hastexo.com/now

> 
> Beginning to wonder if there's either a version mismatch problem between those packages perhaps.
> 
> These are the package versions:
> 
> corosynclib-1.4.1-4.el6_2.1.x86_64
> clusterlib-3.0.12.1-23.el6.x86_64
> pacemaker-cluster-libs-1.1.6-3.el6.x86_64
> resource-agents-3.9.2-7.el6.x86_64
> cluster-glue-libs-1.0.5-2.el6.x86_64
> pacemaker-libs-1.1.6-3.el6.x86_64
> corosync-1.4.1-4.el6_2.1.x86_64
> pacemaker-cli-1.1.6-3.el6.x86_64
> cluster-glue-1.0.5-2.el6.x86_64
> pacemaker-1.1.6-3.el6.x86_64
> 
> Between yesterday and today updated corosync from
> 1.4.1-4.el6 to 1.4.1-4.el6_2.1
> No change however :|
> 
> OS is CentOS 6.2
> 
> Colleague's now compiling a few heartbeat packages instead to see if that gets the job done.
> 
> This being the config for unicast
> 
> compatibility: whitetank
> 
> totem {
>         version: 2
>         secauth: off
>         threads: 0
>         join:   1000
>         consensus: 7500
>         max_messages: 20
>         interface {
>                 ringnumber: 0
>                 bindnetaddr: 10.26.29.0
>                 #mcastaddr: 226.94.1.1
>                 mcastport: 5405
>                 member
>                 {
>                        memberaddr: 10.26.29.238
>                 }
>                 member
>                 {
>                        memberaddr: 10.26.29.239
>                 }
>                 ttl: 3
>         }
>        transport: udpu
>         clear_node_high_bit:    yes
> }
> 
> logging {
>         fileline: off
>         to_stderr: off
>         to_logfile: yes
>         to_syslog: yes
>         logfile: /var/log/cluster/corosync.log
>         debug: on
>         timestamp: on
>         logger_subsys {
>                 subsys: AMF
>                 debug: off
>         }
> }
> 
> amf {
>         mode: disabled
> }
> 
> 
> Any further ideas perhaps? Since I'm losing the plot myself somehow :(
> 
> Cheers!
> 
> Achim
> 
> -----Original Message-----
> From: Regendoerp, Achim [mailto:Achim.Regendoerp at galacoral.com] 
> Sent: 15 March 2012 10:56
> To: The Pacemaker cluster resource manager
> Subject: Re: [Pacemaker] Nodes unable to connect / find each other
> 
> Hi,
> 
> According to the network guy the multicast is all there, but we've got no step further, so we're trying your suggested method now :)
> 
> Didn't know that corosync now supports unicast too, must've missed that...
> 
> Thanks for the heads up!
> 
> Achim
> 
> 
> 
> -----Original Message-----
> From: Andreas Kurz [mailto:andreas at hastexo.com] 
> Sent: 14 March 2012 21:56
> To: pacemaker at oss.clusterlabs.org
> Subject: Re: [Pacemaker] Nodes unable to connect / find each other
> 
> On 03/14/2012 08:22 PM, mark - pacemaker list wrote:
>> Hi,
>>
>> On Wed, Mar 14, 2012 at 1:43 PM, Regendoerp, Achim 
>> <Achim.Regendoerp at galacoral.com 
>> <mailto:Achim.Regendoerp at galacoral.com>>
>> wrote:
>>
>>     Hi,____
>>
>>     __ __
>>
>>     Below is a cut out from the tcpdump run on both boxes. The tcpdump
>>     is the same on both boxes.____
>>
>>     The traffic only appears if I set the bindnetaddr in
>>     /etc/corosync/corosync.conf to the machines' individual IP instead
>>     of to 10.26.29.0 (as advised by howtos).____
>>
>>     Having the latter IP as bindnetaddr, there's no traffic at 
>> all.____
>>
>>     __ __
>>
>>
>>
>> I initially had some communication issues like these, then I realized 
>> I'd forgotten to set up my switch/vlan to handle multicast.  Oops.  
>> With that fixed, everything came up and has worked wonderfully since.
> 
> Or use upnp ... unicasts if you don't want/can change your network setup
> 
> Regards,
> Andreas
> 
> --
> Need help with Pacemaker?
> http://www.hastexo.com/now
> 
>>
>> Something to check, anyhow.
>>
>> Regards,
>> Mark
>>
>>
>> _______________________________________________
>> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org 
>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>
>> Project Home: http://www.clusterlabs.org Getting started: 
>> http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>> Bugs: http://bugs.clusterlabs.org
> 
> 
> 
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
> 
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org
> 
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
> 
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org