Everything related to Maxwell network rendering systems.
User avatar
By joaomourao
#364202
We have several animations on cue for rendering, but the computers we have will have to do the job in time... :(
So in the office we work from 9h to 19h. We are a team of three guys. Each one of us have an eficient workstation to work on. We have two mores computers, one is a Server and the other one used as a Render PC.
What we want is to get to the office in the morning and release all workstations (3) from their rendering process leaving the Server and the Render PC still processing.
I think this should be done easily so that no calculation is lost...
So we kindly ask how can we:

1- Release a node in the middle of the job without losing any calculation, whether it is an animation or cooperative job
2- After releasing the available nodes should pick up this uncompleted job and finish it, whether it is an animation or cooperative job
3- Join in a node and automatically detect which job is incomplete and needs calculation, whether it is an animation or cooperative job
4- No need to connect to the manager for joining jobs. A simple shortcut to a "join any job" button and also a "release from any job" button
5- Add render nodes to job with selected number of threads, whether it is an animation or cooperative job
...no need for remote desktops...

This way we could use all of our Computing power easily even when we are not in the office!
Is this possible right away or are you going to make it happen? :)

Cheers!!
By Kevron
#364213
+1 on these questions. Seems that we share much of the same setup as you guys, so this would be interesting to know how to release a node to be picked up by another node.
By dmeyer
#364265
joaomourao wrote:Hi dmeyer...
So which 3rd party render manager you use? Couldn't it be done my Maxwell network manager?!
Please do try the cooperative part and give us some feedback...
Having computing power without being able to manage it seems dumb for us...
Cheers!
We use Qube from PipelineFX. It is great. There are many others as well though.

I can't really speak to what the current capabilities of Maxwell Network are because we stopped using it some time ago.
User avatar
By joaomourao
#364410
Thanks dmeyer!
I was trying to push NL to perfect their network software by adding some basic functions...
But it seems everyone is avoiding doing animations via network :(
For exemple... you cannot do a batch or animation sequence under network and getting it to resume (mxi) calculations! Only via single render. Try it out :(
I Hope this kind of issues will be resolved in the next release :)
Cheers!
User avatar
By naxos
#370686
dmeyer wrote: We use Qube from PipelineFX. It is great.
I'll have a look at this one...

I do love working with Maxwell (vs VR for ex.), but the network is a pain in the a...
By Gary Bidwell
#370694
I can confirm that cooperative rendering of Maxwell jobs works fine in Qube.

I'd suggest trying the Qube 30 day demo and seeing if it fits your needs. The trial period gives you access to email support so you can iron out any problems you encounter.
By dmeyer
#370698
naxos wrote:
dmeyer wrote: Except for the cooperative part, but we haven't tried.
Will you accept to make a rapid test to tell us if it is ok ?
Sorry forgot about this but it appears Gary has answered.
User avatar
By Bubbaloo
#370711
numerobis wrote:But WHY does it work with an external manager and not with the original one?!?
We render probably 50 cooperative stills (mostly 5760x3740) a week over 12 nodes with no problem using Maxwell's network. I would bet that if a user is having network related problems, it wouldn't matter which render manager software is used.

If you are consistently experiencing problems, you need to isolate the problem, make it repeatable, and that will help Next Limit's team help you figure out how it's failing.
By kami
#377760
Bubbaloo wrote:
numerobis wrote: If you are consistently experiencing problems, you need to isolate the problem, make it repeatable, and that will help Next Limit's team help you figure out how it's failing.
That's what I'm trying to do for quite a while now ... As far as I can tell, I am much happier with v3 as it is much more stable. But still has some unnecessary flaws.
I also considered a 3rd party solution, but they were too work intensive to setup for my taste and quite costly.
As long as you don't try to do too much action with it, it works good now :)
But adding nodes, releasing nodes etc is still buggy or not well thought through

I didn't do animation jobs with v3 yet, but considering your questions, my comments:
1 - I think this almost works now ... If you click on a node and release it, it asks if it should finish the job first (I guess/hope in an animation job if finishes only the current frame). Not 100% what you asked for, but if the frames are not rendering for tooo long, it should work?
2 - I've also asked for that ... sadly it does not work on animations, so you'll manually have to resume the incomplete frames in the end ... a lot of lost work time and risk for errors
3 - why would you need that?
4 - don't understand this question. couldn't you do this with the monitor?
5 - don't understand this question either :(
And why the remote desktop? Couldn't you start and observe the jobs from any machine using the maxwell monitor?

btw, a quick work around I use from time to time, which may solve some problems:
reduce the affinity of the rendering process by a few cores in the taks manager on the workstation. that way you can still work fine and you can just add those cores back again in the evening when you leave the office.

Haha, thanks.

Hello, I'm still waiting for a solution to the pro[…]

Well.....they must have been proven wrong, as it's[…]

Hello dear customers, We have just released a new[…]