DR question

[CCS]

I have a question. Can I include machines in DR that are not in my *local* network? And if so, how?

Obviously, viz/vray would have to be on the slave machine. What if the machine is behind a router? Can I drill thru the ISP-assigned IP address to specify the IP of a particular machine?

This would be very cool if possible

[TishoChaos]

We didn't check that yet, but if it was that easy, I would probably take the IPs of half of the people in the forum and use them to render my images.

[CCS]

heheh. That's why my PCs slowed way down...

Obviously, there would be setup and security issues, but if it were possible I could see application for groups of users to share their processor down time amongst themselves.

[dbuchhofer]

We (#Vray) did some testing of this with the previous implementation of DR.. (the max5 only one..) and it worked, but the transfer time of the file/buckets was prohibitively slow

[CCS]

(1) What internet connection was used in your testing? Just wondering if there would be any point at which it would be feasible.


(2) Similar but different topic:
What about using a similar setup for rendering animation frames?

[RErender]

Backburner works pretty well over the internet, if that's what you meant CCS

[CCS]

Can someone give me some pointers on setting up backburner to render across the internet?

Assume I have several machines on site (local network) and several machines available at each of a couple other sites. Setting aside issues of network speeds, etc. for the moment - how would I get this running? And is it possible to set it up so that people in each of these locations could access the group of computers?

Not sure how you would set up IPs, etc. Mutliple copies of server running to point to each IP? How to get past the IP of the site to the individual machines at the site?

Lots of questions.....

Thx

[RErender]

been awhile since i played with it but make sure the manager is in a dmz with a static IP, then just point the servers to that ip/port, should work :P

[Da_elf]

egz and i had tested the DR waaay back when it was in its arly stages and we were rendering over long distances. problem comes when your trying to log onto someones network of computers. normally the routers IP address is the access from the net. but then it creates its own ips for the networked computer that cant be individually accessed.

[CCS]

I've heard DR is limited this way. I was asking about BB. Does the same apply?



(I've about given up on DR. It's a great idea, and when it works it's awesome. But I spenc more time rubberbanding and duct-taping it than I do using it. So far, BB stripes suit me just fine. Plus, I can keep using my machine while it's running.)

[Da_elf]

does anyone know how to do region renders with BB strips? if i just wanted to render the left side of the image using strips it still renders the whole image. so why the heck does max warn me that ive got it in region mode if it doesnt make a difference anyhow

[czoog]

Well, I've never tried to do exactly what you are thinking of doing Scott, but I have been able access and control/manage the BB renderqueue at my office from home computer. If I had viz at home I am pretty sure i could also submit jobs to the queue, cause I've been able to get it to work with combustion which also uses backburner.

But there can be problems with windows DNS that you should be a aware of:

Here is the way we have it set up in our office:

Our office has 3 subnets 192.168.1.X (workstations, DNS/DHCP sever), 192.168.0.X (all other servers), and 192.168.2.X (rendernodes)

We have a VPN sever that assigns local ip addresses in to remote computers (my home computer) I think in the 192.168.1.150-250 range, but that doesn’t really matter.

So I tunnel in through the VPN, I grab an IP address, etc. Now, this is when I typically have a reoccurring DNS problem that I have never been able to solve without extensive messing around with route tables, etc. My home router deals out IP’s in the 192.168.1.X range, so when my home computer looks to resolve a host name via DNS it looks to 192.168.1.2 (the ip of the office primary DNS server) which it thinks is on my local network, so of course it fails. If I point it to our secondary DNS sever on the 0.X range it works fine, and host names will resolve.

If I use IP addresses that on are a different subnet that my home network, it will always work. I can fire up the render queue monitor, connect to 192.168.2.2 (monkey1, our BB manger) and it works fine

So the long and short of it is, to make it easy on yourself, be sure the rendernodes (BB manger and servers) are on a different subnet than the network you send send jobs from. I’m sure there is an easier to fix to that conflict, but I just haven’t had time to figure it out yet.

If the BB manager and some BB servers are local, and other BB severs are remote, it shouldn’t matter too much, just use static IP’s

good luck!

-Z

[CCS]

ok... I think it makes sense, but will require a few more reads. Does remote backburner always require VPN then?

[czoog]

ok... I think it makes sense, but will require a few more reads. Does remote backburner always require VPN then?

It does sound confusing... sorry. And yes, I would think that you would definetly need a VPN, unless you connect to a single node has a routable IP (unliklwy) or like RErender said, you through it outside the network in a DMZ, but IMHO that's not the safest solution.

I guess you could use port forwarding via the remote router if you only need to connect to a single rendernode per remote location, but that will get messed up the minute thier ISP gives them a new IP address lease.