[ejabberd] about fallback

sergio mailbox at sergio.spb.ru
Sun Apr 17 06:35:46 MSD 2011


Hello.

Does ejabberd clustering allow to make a fallback service, or it's only 
needed
for high load servers and load balancing?
I wasn't able to start one node, when other is down.

OK. Just for test, without synchronize.
I have this dns structure:

_xmpp-client._tcp.my-jabber.tld. SRV 10 0 5222 master.tld.
_xmpp-server._tcp.my-jabber.tld. SRV 10 0 5269 master.tld.

_xmpp-client._tcp.my-jabber.tld. SRV 20 0 5222 slave.tld.
_xmpp-server._tcp.my-jabber.tld. SRV 20 0 5269 slave.tld.

master.tld. A ip.address
slave.tls.  A another.ip.address

I've copied mnesia via backup to slave, stop ejabberd on master, and run 
ejabberd
on slave. All works fine. But I've problems with gmail.com and 
jab.undernet.cz:

I(<>:ejabberd_s2s_out:353) : Closing s2s connection: my-jabber.tld -> 
jab.undernet.cz (invalid dialback key)

I(<>:ejabberd_s2s_out:417) : wait for validation: my-jabber.tld -> 
gmail.com (closed)


Loglevel 5 doesn't help:

for jab.undernet.cz:
D(<0.881.0>:ejabberd_s2s_out:197) : open_socket: {"my-jabber.tld", 
"jab.undernet.cz", "2778895764", false}
D(<0.881.0>:ejabberd_s2s_out:1022) : srv lookup of 'jab.undernet.cz' 
failed: nxdomain
D(<0.881.0>:ejabberd_s2s_out:1107) : inet of jab.undernet.cz resolved 
to: [{81,2,197,210}]
D(<0.881.0>:ejabberd_s2s_out:268) : s2s_out: connecting to 
{81,2,197,210}:5269
D(<0.881.0>:ejabberd_s2s_out:340) : recv result: 
{"jab.undernet.cz","my-jabber.tld",[],[]}
I(<0.881.0>:ejabberd_s2s_out:353) : Closing s2s connection: 
my-jabber.tld -> jab.undernet.cz (invalid dialback key)

for gmail.com:
D(<0.1221.0>:ejabberd_s2s_out:268) : s2s_out: connecting to 
{74,125,77,125}:5269 (xmpp-server.l.google.com)
I(<0.1221.0>:ejabberd_s2s_out:417) : wait for validation: my-jabber.tld 
-> gmail.com (closed)


-- 
sergio.


More information about the ejabberd mailing list