Silly Multicast Question (Queue)
-
In previous versions of FOG, there was a limit (10 by default) where when you multicast for active PCs. Now, using 1.1.2, is that setting or default changed? I had a group with 32 computers in a lab and started them today assuming only specified number (figured 10 as default) would start, but all of them started. Sadly I have 10/100 switches to the class rooms and a Gb backbone. Has this changed or am I just missing the setting?
I figured something like this would be displayed when I started my multicast:
[url]http://www.fogproject.org/wiki/images/3/3d/Queue.jpg[/url]
Edit: I guess reading the wiki further states this was only for [I]ten[/I] unicast tasks, but multicast is considered one task itself. Is there a way to break the multicast down into parts other than making a separate group? I don’t know why 32 multicasts would be significantly slower than 8.
-
[quote=“RLane, post: 32598, member: 23505”]In previous versions of FOG, there was a limit (10 by default) where when you multicast for active PCs. Now, using 1.1.2, is that setting or default changed? I had a group with 32 computers in a lab and started them today assuming only specified number (figured 10 as default) would start, but all of them started. Sadly I have 10/100 switches to the class rooms and a Gb backbone. Has this changed or am I just missing the setting?
I figured something like this would be displayed when I started my multicast:
[url]http://www.fogproject.org/wiki/images/3/3d/Queue.jpg[/url]
Edit: I guess reading the wiki further states this was only for [I]ten[/I] unicast tasks, but multicast is considered one task itself. Is there a way to break the multicast down into parts other than making a separate group? I don’t know why 32 multicasts would be significantly slower than 8.[/quote]
Sounds like you figured it out there towards the end, Unicast has a limit to how many machines can be imaged at once, it can be increased or decreased by modifying the configuration settings.
As far as Multicast goes, your best bet is to start a group and name is Session1, Session2, etc. and put 10 or 12 machines in there and run the multicast for the group.
Unfortunately at the moment there are no ways to break down a multitask imaging task any further, but it may be something we will look into in the future.
Personally, I prefer to unicast to multiple machines at once, as you stated there is already a 10 client limit, and you can set up multiple machines to image and when a space opens in the queue it will fill it otherwise it will wait. This may be a little more beneficial to you than multicasting since you seem to want your machines to wait their turn so not to bog down the network, I’ve had multicast hurt me more times than it has helped.
If you want to use the Queue you need to use Unicast. Multicast makes sure all hosts are on the same “checkpoint” before proceeding and does not rely on the Queue.
