if you are using multicast set the metadata_interval to 0. I suspect you
are causing a multicast storm which is causing things to die off.
On Wed, 9 Nov 2011, Avishai Ish-Shalom wrote:
Actually i've set it to 10. Also, tcpdump on a few nodes revealed that
multicast messages are being sent but no messages are received.
if i restart gmond then on that node i see messages being recieved for a few
minutes before the problem comes back, while on other nodes
no messages come up on tcpdump.
My best guess is that we are dealing with some igmp issue and that restarting
gmond rejoins the multicast.
Regards,
Avishai
On 09/11/11 21:33, Rick Cobb wrote:
Just offhand and without doing my googling, this sounds like you haven't modified
the gmond "send metadata" timer (sorry,
name's not handy). Vladimir's cheat sheet is pretty clear about that: if
you're unicasting, you have to modify that
parameter.
-- ReC
On Wed, Nov 9, 2011 at 7:23 AM, Avishai Ish-Shalom <avis...@fewbytes.com> wrote:
I'm having the exact same problem, my configuration is similar. ganglia
3.2.0 installed from lucid ppa, network is one layer 2 segment spanning
3 switches. I've tried downgrading to 3.1.2 on one node and got the same
behaviour.
Anything I can do to help debug this?
--
Regards,
Avishai
------------------------------------------------------------------------------
RSA(R) Conference 2012
Save $700 by Nov 18
Register now
http://p.sf.net/sfu/rsa-sfdev2dev1
_______________________________________________
Ganglia-general mailing list
Ganglia-general@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ganglia-general
------------------------------------------------------------------------------
RSA(R) Conference 2012
Save $700 by Nov 18
Register now
http://p.sf.net/sfu/rsa-sfdev2dev1
_______________________________________________
Ganglia-general mailing list
Ganglia-general@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ganglia-general