Thx.
The PullRequest works like the brende commandline did.
But with the same code I couldn’t merge the individual tiles from the subframe rendering. Does anyone know how to merge them? I have visible lines at the boarders.
I would take the role as collaborator. I know how it is to have less time. And it would really be a shame if brenda-web gets “old”.
Next week I’m in an Angular Code Camp and I will try to port brenda-web to angular2 or now just angular 
Hi,
Thanks for putting this together. I’ve been able to set everything up but after letting two xlarge instances run for an hour (500mb file size single frame render) there was still no “combined task time” or “estimated time left” shown. Also CPU load was 0 for both. The status was shown as running and the file was stored on AMS. Your tutorial states the progress should be shown so I assume that nothing was actually happening?
Two possible things:
I could not create a 'myproject.tar.gz" file, everytime it became myproject.blender.gz. I am not familiar with coding but I tried what you had posted and everytime it ended up with blender.gz. Is there something I am doing wrong? I am using mac OS 10.10.5
On brenda-web I have the CORS config error which I tried to update on AMS many times but it doesn’t go away. Could this be the problem?
Also, I had set up my AWS for renderbot before where it was recommended to stick to us-east-1c due to data transfer costs. However that instance, at the moment at least, is the typically the most expensive (by a huge margin sometimes). When using brenda-web does it make a difference which instance is selected as far as addtional costs?
Any help would be greatly appreciated!
Thanks
Has anyone had any luck getting this working lately with 2.79? I noticed there is an AMI for 2.79. All my attempts have failed. Once the instance is launched the CPU load is at 0 so I guess something isn’t working.
Edit: Ok I got this working. Make sure you dont have any capital letters in your buckets. Thats what was stopping it before.
I do have an issue. The CPU load is only 4% which cant be right. The tasks are being completed very slowly. On the EC2 console under monitoring it says %100 cpu load, but these tasks are completeing as if it was only using 4% CPU. I used a c4.xlarge and the render is only 10 samples so it should be cutting through frames.
Edit for the edit: I tried trimming my project down to only the essentials. It renders on my I7-6700k at 10 samples in 2 minutes. I now tried t on the C4.8Xlarge which is a lot more cores and its taking 20 min a frame and it says the CPU usage is at 35%. What is going on here?
Ok so, for some reason it renders a lot quicker on the 2.78 ami, it still shows only a cpu load of 1.9% so I dont know how accurate that thing is. What would be causing such a significant change in render times between the two blender versions? And how can I make sure the instance is using %100 of the cpu or at least as close to it as possible.
Edit: I’m pretty sure the cpu load is related to the VCPU’s not percentage. So it is running at %100 but it doesn’t explain why it is so damn slow for 2.79. I will render out some test images to compare.
I have an AMI with 2.79 running without problems. I don´t want to make it public because the brenda git website notes that you should
be sure to clean the instance filesystem of security-related files before you snapshot it but I have no idea how to check if there are security issues.
A frame taking 20 minutes on an c4.8xlarge which renders in 2 minutes on your own pc can´t be right.
Check the log for problems with:
brenda-tool ssh tail log
Ok got this working. It is really awesome, I hope development continues on it. The issue must have been something to do with my project file. I trimmed the fat from it and did a test render at 1000 samples instead of 10. On my machine it took 90 minutes on one c4.8xlarge it took 30 minutes.
I want to thank you Lumpengnom for putting up with my lack of Linux know how on this thread and the renderbot one. You helped a lot! And I learned some new things.
Good to hear that it finally worked! 
Note that if you want to render with more than one c4.8xlarge you have to request a limit increase as on new accounts these instance types (among others) are limited to one instance.
If you want to use more than (i think) 20 instances in total you also have to increase your maximum total limit.
You do that in the EC2 console under “Limits” in the left hand menu.
So far I have found the c4.8xlarge the best instance in terms of speed vs. power but I havent´t tried the new c5.9xlarge yet which have quite some more cores but also come at a higher price.
I’ve used renderbot last year for a project and requested an instance increase so thats all good. C4.8xlarge is the most cost efficient. The c5’s are not available on Brenda web it looks like. Rendering with GPU is more expensive than its worth in regards to rendering cycles. Sure it gets more frames done in an hour but not enough to offset the price increase.
First off, apologies for the delay in response, it has been busy on my end.
You are correct, the CPU load figure that appears in the dashboard is the standard way in which linux reports load and is related to the number of VCPUs the EC2 instance has. Some details on how to interpret this figure can be found here https://en.wikipedia.org/wiki/Load_(computing) but at a high level: if your EC2 instance has 2 VCPUs then the maximum it’s load can be is 2.0. For example a load of 0.5 on a 2 VCPU instance would represent 25% usage, 1.0 would be 50%, and 2.0 would be 100%.
I updated the instance list and it should now includes the c5 instances as well.
Do the c5 instances need some modifications of AMI or work staight out of the box? I use c4.8xlarge vm-s with brenda scripts through Ubuntu and haven’t tried the c5-s because somehow I got the impression that older AMIs don’t work on them.
I haven’t tested with the c5 instances but the idea behind using AMIs is that they should be able to run on any instance type. I can try in the next few days and post what I find unless someone beats me to it.
I tried the c5.18xlarge and they worked fine with a 2.78 AMI.
I am using this again for a new project but lately EC2 has given me a “Capacity-Not-Available” error when it comes to the spot requests. I am bidding over the price so I am not sure why I am getting this error. Could someone else check to see if they are getting it on their account?
I am requesting 1 instance of c4.8xlarge on us-east-1. It took me 6 hours to get an instance running. Earlier this year it would go instantly if my bid was above the spot price.
I want to make sure this isn’t some issue with just my account.
Ok I got some answers here.
Looks like the way EC2 operates spot pricing has changed. I preferred the old way thats for sure.
http://autoscalr.com/2018/01/04/new-aws-spot-pricing-model-good-bad-ugly/
hi guys! I am trying to launch a job with brenda web but there are no AMI presets available. I can’t understand if I need to manually install something on the instances or if it’s enough to use a preset for the official 2.79. Where can I find the AMI presets?
other question. I saw that it’s possible eventually to install a custom version of blender on the instances and then create a custom AMI. Is it possible to automate such process on several instances or would I have to do it manually on every instance I wanna render on?
Thanks in advance for any help! 