Multicast ( again ) .... imaging stops while processing

Flavalf · May 31, 2017, 2:49 PM

Server

FOG Version: 1.4.1
OS: Debian 8

Client

Service Version:
OS:

Description

So, after make my network works, I face a new problem …
I create a Multicast task, it starts on all clients, but after a few minutes, the imaging process stops !
As you see in the multicast.log, the task ends itself :

[05-31-17 10:39:10 pm] * No tasks found!
[05-31-17 10:39:00 pm] * No tasks found!
[05-31-17 10:38:50 pm] * No tasks found!
[05-31-17 10:38:40 pm] * No tasks found!
[05-31-17 10:38:30 pm] | Task (2) Multi-Cast Task has been completed.
[05-31-17 10:38:30 pm] | Task (2) Multi-Cast Task is no longer running.
[05-31-17 10:38:20 pm] | Task (2) Multi-Cast Task is already running with pid: 21947.
[05-31-17 10:38:10 pm] | Task (2) Multi-Cast Task is already running with pid: 21947.
[05-31-17 10:38:00 pm] | Task (2) Multi-Cast Task is already running with pid: 21947.
[05-31-17 10:37:50 pm] | Task (2) Multi-Cast Task is already running with pid: 21947.
[05-31-17 10:37:40 pm] | Task (2) Multi-Cast Task is already running with pid: 21947.
[05-31-17 10:37:30 pm] | Task (2) Multi-Cast Task is already running with pid: 21947.
[05-31-17 10:37:20 pm] | Task (2) Multi-Cast Task is already running with pid: 21947.
[05-31-17 10:37:10 pm] | Task (2) Multi-Cast Task is already running with pid: 21947.
[05-31-17 10:37:00 pm] | Task (2) Multi-Cast Task is already running with pid: 21947.
............
[05-31-17 10:30:20 pm] | Task (2) Multi-Cast Task is already running with pid: 21947.
[05-31-17 10:30:10 pm] | Task (2) Multi-Cast Task has started!
[05-31-17 10:30:10 pm] | Command: /usr/local/sbin/udp-sender --interface eth0 --min-receivers 1 --max-wait 240 --portbase 53350 --full-duplex --ttl 32 --nokbd --nopointopoint --file /images/W7/d1p1.img;/usr/local/sbin/udp-sender --interface eth0 --min-receivers 1 --max-wait 10 --portbase 53350 --full-duplex --ttl 32 --nokbd --nopointopoint --file /images/W7/d1p2.img;
[05-31-17 10:30:10 pm] | Task (2) Multi-Cast Task sending on base port: 53350.
[05-31-17 10:30:10 pm] | Task (2) Multi-Cast Task 1 client found.
[05-31-17 10:30:10 pm] | Task (2) /images/W7 image file found.
[05-31-17 10:30:10 pm] | Task (2) Multi-Cast Task is new!

And clients hang in imaging process …

This issue appears at each time, with 3 or just one client …

Flavalf · Jun 1, 2017, 6:50 PM

Good news everyone !

One more time, FOG was not faulty but the NETWORK …
After checking again Forums on multicasting and similar problems, I’ve found the solution :

The querier switch must have an IP adress on the subnet …

Once I’ve made the change, I tested a 4 client multicast session : total success !
So I tried with a classroom of 19 clients : in 30 min ( 1.5Gbps/min, max speed with 100Mbits connections ), all images were downloaded !

So, maybe it would be interesting for others to include these advices in the wiki or the troubleshooting Multicast page :

enable IGMP snooping on VLAN concerned in all switches

Choose 1 "master switch ( for me the all gigabit one under fogserver ) and make the following only for this one :

assign an IP adress to the VLAN in the switch configuration in the subnet range
enable IGMP query on the VLAN
verify the switch has Active Group Addresses
Enjoy !!!

Thanks again at all the FOG team

Tom Elliott · May 31, 2017, 8:56 PM

Perhaps there’s a data gap? Meaning your network stops allowing traffic through? I don’t really know.

The multicast logs would be extremely helpful here. Essentially there’s two multicast logs, one with the main information (which you sort of presented) and another that’s the pure data. I’m not seeing a reason for the imaging to just stop though.

Flavalf · May 31, 2017, 9:10 PM

Thank you Tom, where can I find the other multicast log ?

Sebastian Roth · Jun 1, 2017, 8:28 AM

@Flavalf This sounds like a tough issue to find out about. Probably best if you could the udpcast tools by hand to see the exact messages. From the log file you posted you already have the command you need to run in the server side. Then boot up one of your clients in debug mode (FOG web interface -> create task -> deploy, but check debug). When you get to the shell run the following command. Hope I got all the parameters right:

udp-receiver --nokbd --portbase 53350 --file /dev/null

Flavalf · Jun 1, 2017, 6:50 PM