SVN 3710 : udp-sender send multicast to eth0 instead of em1
-
Hello,
after creating topics in the wrong way, here is the a bug : in multicask session, the udp-sender command send multicast stream to network interface eth0 instead of em1 in my case :
==> multicast.log.udpcast.28 <==
The following interfaces are available:
lo 127.0.0.1
em1 172.31.43.1
cat: erreur d’écriture: Relais brisé (pipe)
Udp-sender 20120424
No suitable network interface found
The following interfaces are available:
lo 127.0.0.1
em1 172.31.43.1
cat: erreur d’écriture: Relais brisé (pipe)Settings are correct in fog settings.
-
Seems the getInterface() function return the bad one…
-
The get interface method used to get the udp cast interface, which works only if there is one master node in your network. And that the interface is set to get the proper names interface. However each case node has its own interface set as well. I set the get interface method to grab the images group master nodes interface so the image can be multicast end properly regardless of where the image is located. Check your storage nodes to ensure the interface for that node is correct. Trunk should work normal now because it detects the real interface name and not always assume eth0 will be the name. Hopefully this makes sense.
-
I am running git version 4196 of fog on a red hat release 6.6 system. I can no longer multicast it just hangs on the clients waiting for the multicast to start. The multicast.log file shows that the multicast server is picking up eth1 instead of eth0 for the multicast interface (both are active). The FOG_UDPCAST_INTERFACE is set to eth0. Is there someplace else to look?
-
@tmerrick update your version please and try again. This time pay attention to the master nodes’ interface settings as this is now what multicast manager is using to send out the tasking.
-
At 4236 and it is still going to eth1. There is only one Storage node and it has eth0 defined in its interface. Is the master interface setting set someplace else?
-
It is fixed at 4250.
