PDA

View Full Version : SquidNet on Mac OSX for LightWave



Rollie Hudson
04-03-2017, 09:48 AM
Help! Does anyone have any experience with SquidNet render manager and LightWave 2015.3 using Mac OSX? We've been trying to set up a small farm - so far unsuccessfully. We've got 8 MacPro towers networked in our studio - (using a Netgear 16 port gigabit switch - which is decent but which actually does NOT have link aggregation). On the networking front, all of the computers are mapped properly et al. We're starting off with a simple scene in LightWave - not even using any plugins at this point - with all content in the same directory. But we just can't seem to get SquidNet to take off and fully render the scene. We keep getting error messages in SquidNet without any further explanation. Next, we're going to try doing full installs of LightWave on each machine but I'm not sure that's going to solve anything, as, for example, ScreamerNet alone only needs the small macro installed on each machine. Any advice @ network rendering for LightWave in the OSX environment would be helpful.

Scazzino
04-03-2017, 10:50 AM
Don't know anything about SquidNet, but give the free version of my DreamLight Constellation (http://dreamlight.com/shop/dreamlight-constellation-lightwave-3d-network-render-controller/) a try. It was specifically designed for a Mac based small render farm using Mac drag-and-drop simplicity for my own Mac based studio.

BTW: I see you had purchased DLI_SNUB-Launcher in the past. DLI_SNUB-Launcher is the GUI utility to configure and launch the LWSN nodes themselves. DreamLight Constellation is the companion product that is a full network render controller that works with DLI_SNUB-Launcher and LWSN.

Rollie Hudson
04-26-2017, 04:21 PM
We've managed to get SquidNet working with all the render nodes. Previously, a job would only process on one node, even though a pool of nodes was available. This seems to be because the project can't be split into more than one "slice". No matter what settings are entered as far as frames per slice, etc. the project only renders on one node, as one slice encompassing all the frames.
I have found a workaround that does allow rendering on all nodes simultaneously. Basically, we create an individual "job" for a specific set of frames, (i.e. 1-15 or 16-30), then assign that job to only process on a specific node. All settings for content directory, config folder, output directory, are constant. It works, but the setup is extremely tedious.
This problem does not present when working with any other template in SquidNet-- just LightWave.
Does anyone have any idea why LightWave in SquidNet will only render on a single node? Our workaround is functional, but we would love to not have to manually split up each job into slices and assign to each individual node.
Here's some specifics about the setup:
All nodes (and the controller) access a network drive containing the installation of LightWave and all content and output directories. The drive mounts as /Volumes/VILLAGE. We checked in terminal, and the drive is mapped correctly on all nodes; (not mounted as VILLAGE -1). All nodes and the controller access the drive with the same username and credentials, in an effort to negate any permissions issues. Just for clarification, "VILLAGE" is a raid volume that is hosted on the same machine that controls SquidNet. It's been recommended that we use a NAS so that the path is constant across the board, but since it mounts on the local machine and the nodes as /Volumes/VILLAGE, there's no problem.

http://www.squidnetsoftware.com/

Rollie Hudson
04-26-2017, 04:27 PM
Thanks for your response here, Michael. I've got your book and yes, I'm a customer of yours for the DLI_SNUB. However, I work with an IT associate who knows considerably more about drive mapping etc. than I do. I, on the other hand, know more about LightWave, having used it for many years. At this point, with the particular set up of MacPro towers we have, we were not able to figure out how to get them rendering together using DLI-SNUB, probably simply due to our own inability to navigate the intricate IT set up needed to get your render controller working, not due any fault with your software. However, I see that you've just released 2.0. If we're not able to get SquidNet running - or even if we are - we may end up purchasing your update and having a go getting the render farm going with that. Keep up the great work and thanks again for checking in.

3dworks
04-27-2017, 05:47 AM
hi rolle, i did use squidnet in the past, but nowadays i'm rendering only with octane, which has it's own distributed rendering inbuilt. i remember ist was a bit tricky to set squidnet up but in the end it did its job very well. this was over 2 years ago, in the meantime the versions of squidnet and OSX progressed and on OSX side many things changed - also i don't really remember the little tricks and tips, but the programmer, ralph alvarez, was extremely friendly and helpful getting back over skype to fix all OSX issues. did you try to contact him? i will send you the skype contact as private message...

cheers

markus

Rollie Hudson
04-27-2017, 10:27 AM
Again, thanks Markus, for the suggestion and info @ Ralph... We are also rendering with Octane and LightWave, but as LightWave 2015.3, and LightWave Next are biased, CPU-based renderers as I understand it, we're still putting time and effort into getting these 8 MacPro workstations functioning as a renderfarm. It's great to have both options with LightWave. Hopefully, we'll see LightWave Next soon and it's renderer will work on this mini-farm set up.

Scazzino
04-27-2017, 07:28 PM
Glad you're up and running Rollie!