Dr + bb

[glorybound]

I never use BackBurner, because I don't really trust it, and I don't really understand both working together. Just to clarify. Is it good to run BB and DR together? I usually run DR on my main workstation, and all my slaves, but how does it work with BB?. It seems the slaves are working away on the whole image, on my workstation and the slices across the network, which doesn't seem correct. Please, can someone explain the best practice?

[cheerioboy]

I have backburner servers and DR spawners running on every render node, you just have to be careful not to run two jobs on the same machine simultaneously. It can be easy to forget that one job is using the other nodes for DR, and you send another job though backburner to those nodes.

One option is to send an animation to render without DR, having single frames picked up by each machine. The other option would be to send the whole job to one machine, but have DR turned on so that it connects with the remaining nodes to assist on each frame through DR.

I'm doing stills these days, so I'll have one node be my master which I send the job to, but DR will be ticked with the remaining nodes picked to assist. So that master node is essentially playing the role of the main workstation in your situation. But in this case you can now continue working while the image renders...

Does that make sense?

[PIXELBOX_SRO]

i ll describe how i use it:

I have a main work station where i work and from which i send all the jobs to BB.
On one slave machine i have MANAGER, MONITOR and SERVER running.
On the other slave machines i only have DR running (unless i do animation)

When I lounch job from my main workstation i make sure i have DR on, sending it to the SERVER which is connected to the MANAGER and MONITOR.
THis way, i can send hundreds of jobs if i like and they will be launched one after another only on that SERVER machine, but in DR mode, using all the DR nodes you actually have.

The only thing you have to make sure: dont run your spawner on the machine where you run the server (produces errors)...you dont have to anyway as that machine gets picked up automatically.

Make sure you SERVER machine is the one with the most cores and memory because LC is computed always locally.

That should do it

Try it only a small job that renders at least 10 minutes so you can see if all nodes are being picked up well.

Hope this helps

[PIXELBOX_SRO]

using DR for animation is a NO NO...
youre loosing quite a lot of steam ...i d say 15-25% of computing power.
For animation its best just to use BB without DR, while running SERVERs on each machine an selecting USE ALL SERVERS while submiting the job.

[glorybound]

When I send a job, which are stills, it's to all the machines and it is overnight. Recently, I have been running out of RAM, so I want to use BB to slice it up. What I want is one slice at a time, being processed by all resources, and then it'll move to the next slice. So, essentially, I might have 1 scene sliced into 10, and rendered one after another. BB does what I want, but I am not sure the roll of DR. Would it be as simple as closing MAX on my main workstation and launching it as a slave?

[cheerioboy]

I've never sliced up a rendering into smaller parts to be split up. If there are issues with RAM, couldn't you turn on dynamic geometry and limit the amount of RAM used so that it swaps instead of maxing out and crashing?

[glorybound]

My best machine is my main workstation, so it is running server, manager, and monitor. All the slaves are running the server. I am going to launch a render right now, and show you all the steps. Maybe someone else is also missing out on the potential out of ignorance, like me.

[Donald2B]

Unless I have a MAJOR deadline looming and the still is going to take too long, that is the ONLY time I ever use DR.

We are about 90/10 for animation/stills.

My file server runs the Manager and all my slaves run the server and nothing else.
Otherwise, another employee will look in the monitor and see the renderfarm as free, sending a huge animation to the farm, crashing the DR already going. So we run one or the other, but never both at the same time.

We about 40 nodes now, we are looking into something like REFAMO to help switch between the two. We try to be good about planning ahead, especially on the big images.

In the off times when we get a surprise, I leave 4 backburner nodes running and make sure each node is only using 8-10 other nodes and then send a bunch of images to backburner, usually all servers. The 4 nodes will pick up the images and then use the DR servers setup on each of the BB machines and all is good.

[glorybound]

So, you use backburner instead of distribution render? I have 6 slaves, and several of them are dogs, so would clog up things. If I only used BB, those slow machines would take forever, when BR distributes the buckets.

[Donald2B]

For animation we only use BB.

If we have to use DR, I still go through backburner to manage the stills.

So in your case:
- Main workstation is server
- DR servers you want to use are setup on the workstation
- Slaves are running appropriate VRay DR Spawner
- Submit still to Main workstation in BB
- Main workstation will DR render the still and use the other machines as slaves
- submit more stills and BB is now a que for additional DR renders
- go to sleep and profit!

[glorybound]

I like using slices on my main, but DR on the others. I would just remove the others from the manager, correct, leaving my main? My main is almost done with one slice and the slaves are still in LC.

[Donald2B]

Okay, so in this case, you need to delegate one of your slower machines as the DR host computer. Setup the DR server IPs on that computer and then send all of the DR stills to that computer. Then when it picks up the job in BB, it will use the DR computers you specified to render the still.

Just make sure you don't have your main computer in that IP list on the DR Host computer.

So only two computers would be visible in the BB Monitor: Main workstation and DR Host. All the others will be absent since they will not be running BB Server.

[glorybound]

Now I know why I haven't done this yet... what? QUOTE=Donald2B;596146]Okay, so in this case, you need to delegate one of your slower machines as the DR host computer. Setup the DR server IPs on that computer and then send all of the DR stills to that computer. Then when it picks up the job in BB, it will use the DR computers you specified to render the still.

Just make sure you don't have your main computer in that IP list on the DR Host computer.

So only two computers would be visible in the BB Monitor: Main workstation and DR Host. All the others will be absent since they will not be running BB Server.[/QUOTE]

[glorybound]

My slaves don't have licensed MAX, so how do I change the DR Server IP's? Is this the V-Ray license server information settings? It's a different V-Ray install, isn't it?

[Donald2B]

C:\Users\**UserName**\AppData\Local\Autodesk\3dsMa x\**MaxVersion**\ENU\en-US\plugcfg

**UserName** = user name you are working under (Troy - in my case)
**MaxVersion** = the version of 3DS Max you are using (2014 - 64bit - in my case)

That is the location of the vray_dr.cfg file.

Just copy this file from your workstation and paste it into the same location on the DR Host computer running BB Server.

Edit the file and change the 1/0 at the end of each IP address / computer name. 0 = Off 1 = On

Now you should be able to send a BB job to that BB Server with DR enabled in VRay. Then the BB Server will then know which machines it can use for DR.

Makes perfect sense!