Thanks for the update Achim.
I'm discussing with Hazelcast guys to understand why they changed this
and their plan.
Regards
JB
On 01/21/2015 08:26 PM, Achim Nierbeck wrote:
Ok, I'll revert my -1 to a not so happy +1 (binding)
after changing the hazelcast file it also works on mac
regards, Achim
2015-01-21 10:28 GMT+01:00 Jean-Baptiste Onofré <[email protected]>:
Good point, I will add a note in the documentation.
Regards
JB
On 01/21/2015 09:42 AM, Achim Nierbeck wrote:
Hi,
I'll only be able to test this later this evening. Never the less this
will
be an issue for a couple people then.
We need to have a hint for that at least.
Regards, Achim
sent from mobile device
Am 21.01.2015 09:11 schrieb "Jean-Baptiste Onofré" <[email protected]>:
Yes, but previously we use both: multicast AND tcp-ip. So even if it
didn't work in multicast, it uses tcp-ip. So maybe your interface never
worked with multicast ;)
Now, this config is no more possible: it's multicast OR tcp-ip.
Can you test the tcp-ip config ?
Regards
JB
On 01/21/2015 09:05 AM, Achim Nierbeck wrote:
Strange, I didn't change anything on my Mac network interface. So we
need
to file a bug at hazelcast then?
Regards, Achim
sent from mobile device
Am 21.01.2015 09:02 schrieb "Jean-Baptiste Onofré" <[email protected]>:
yes and no: it's a change in Hazelcast. So what would you propose:
1/ disabling multicast and enable tcp-ip: in that case, it's a
regression
for user using multicast on different machines. It will work in your
case
for node on a single machine without multicast support on the interface
(by
the way, it's the config used in hazelcast module tests as the tests
run
on
a single machine)
2/ enabling multicast and disable tcp-ip: it's the current case, it
works
for users on different machines using multicast, not when multicast is
not
allowed.
So ...
I would prefer the current situation (which works if your network
interface supports multicast).
Regards
JB
On 01/21/2015 08:51 AM, Achim Nierbeck wrote:
Hi,
As it used to work without this tweak before it's still a regression.
:-)
Regards, Achim
sent from mobile device
Am 21.01.2015 07:36 schrieb "Jean-Baptiste Onofré" <[email protected]>:
It seems that I reproduce the same behavior by disable
promisc/multicast
on my network interface. Just disabling multicast and enabling tcp-ip
in
etc/hazelcast.xml fixes the problem.
Regards
JB
On 01/21/2015 07:21 AM, Jean-Baptiste Onofré wrote:
Ahhhh I may know.
In previous Hazelcast version, it was possible to enable multicast
AND
tcp-ip for joining. Since Hazelcast 3.3.x, it's no more possible.
By default, in the Cellar etc/hazelcast.xml, I enabled multicast and
disable tcp-ip.
If your interface or network doesn't support multicast, it will fail
to
discover the nodes.
Can you try to change to tcp-ip (with localhost) instead of
multicast
?
Thanks,
Regards
JB
On 01/21/2015 12:15 AM, Achim Nierbeck wrote:
Config seed port is 5701 and cluster size is 1. Some of the ports
seem
occupied!
--
Jean-Baptiste Onofré
[email protected]
http://blog.nanthrax.net
Talend - http://www.talend.com
--
Jean-Baptiste Onofré
[email protected]
http://blog.nanthrax.net
Talend - http://www.talend.com
--
Jean-Baptiste Onofré
[email protected]
http://blog.nanthrax.net
Talend - http://www.talend.com
--
Jean-Baptiste Onofré
[email protected]
http://blog.nanthrax.net
Talend - http://www.talend.com
--
Jean-Baptiste Onofré
[email protected]
http://blog.nanthrax.net
Talend - http://www.talend.com
