This is strange. I’ll take a look at it when I get the chance. Thanks for the workaround in the interim.
It appears that the workaround doesn´t work after all. I just noticed that the particle system was rendered as if the system had allways been in the state of frame 1. The camera did move but the particles remained at their original place. Because of the cameras animation I didn´t notice this right away.
Anyway, I think I found out why rendering from a zip file didn´t work on c4 instances. When rendering from the zip file the zip file is first extracted on the instance. The c4 instances seem to have less hard disk space and simply report an out of disk space error after a while.
The discription on the AWS page tells us that c3.8xlarge instnce have 2 x 320GB SSD drives. But in the discription of the c4 instances it says EBS only. The way I understand it EBS (Elastic Block Storage) is storage that is virtually created on demand in the AWS cloud and then attached to the c4 instances.
I assume that AWS or Renderbot somehow sets a size for the volume so that it is enough for the zip file but unfortunately not enough for the whole unpacked zip file or something like that.
Here are links to the instances and the EBS:
There´s another thing: Sometimes specific instance types won´t launch. Renderbot will retrun an error message:
“An unfortunate error appears to have occured. Please double-check your settings and the last action you attempted.”
This happens usually when the price for c3 instances is lowest in us-east-1e.
I tried to launch a spot request for a c3.4xlarge directly from the AWS console and the request guide presents me the price for instances in 1a, 1b, 1d and 1e. But later on I´m asked to chose a subnet from a drop down menu and 1e is not available there. I can only chose from 1a, 1b and 1d.
I read somewhere that 1e is the newest availabilty zone and since c3 instances are older types of instances i´m guessing that there might not be any c3 instances in that region at all and that displaying the price is a bug.
So, would it be possible to make Renderbot so that in case of such an error the use can manually select an availabilty zone?
Hi I’m having trouble with renderbot. I have everything installed and i can log into the terminal but once I hit enter to edit a field I cant ever use any of the navigation numbers again. I can either type or hit enter and go down another line but never leave whatever menu I’m on. Any ideas?
Do you mean this:
I see. I’ll try making the next version a larger size. Up till now I had been keeping the “footprint” of the Renderbot images to a minimum to make them launch as fast as possible, but I’ll try adding a few spare gigabytes to see if that solves this issue.
Yes, I think that should be possible. It would be relatively easy to prompt the user to input a letter (a, b, c, d, or e) to manually select the availability zone.
Yes, Lumpengnom also raised this issue initially. I, however, have been unable to replicate it. I would be happy to take a look at what is going on if you could provide me with the exact sequence of steps you are performing, so I can replicate the error.
well it does it everytime I press “1” for setup. I then press enter to edit the first field. After I edit the field I press enter again and the cursor drops to the next line. I’m assuming I should be able to use the navigation again but it only types the number into the field. I can navigate the menus as long as i don’t edit any fields. I can press enter until I’m off the screen then it goes black.
One thing I noticed and I only bring it up because your instructions were so clear and detailed and I didn’t see these steps included. When I subscribed to the client I had to start a VPC before i was able to continue. Then I had to update the settings to be able to get a public IP. I’m pretty new to all of this. Hope some of this helps. Thanks!
The prices and instances don’t show up either. All the boxes are blank. I’m using Renderbot Client 1.0.4, and Renderbot Node 1.0.3.
You start Renderbot.
You press 1 for Setup.
You are now in the Setup menu.
You press 1 for editing.
You are now in line 1.
You edit line 1 and press enter.
You are now in line 2.
You edit line 2 and press enter.
You are now in line 3.
And so on.
You have to complete all lines before being able to navigate again.
If you don´t want to change a line simply press enter.
What exactly do you do?
You should press 1 for Input instance type.
Then enter an instance type and press enter.
Instance types are for example:
c4.2xlarge
c4.4xlarge
c4.8xlarge
c3.4xlarge
g2.2xlarge
The instance types and what they are best for are listed here:
personally I use mainly
c4.8xlarge
c4.4xlarge
If you want GPU rendering (which is usually more expensive than throwing a hoard of cheaper CPU insances at it) use:
g2.4xlarge.
Note that you have a limit of instances that Amazon allows you to use. In the AWS console under EC2 you can klick a link labled “Limits”. If you need more instances of a specific type you can request a limit increase there.
Hey so I need some help. I believe I had set everything right. Everything appears connected but when I look at my instance cpu usage it is hardly being used so its not rendering. I created logs but i really dont understand what it means. Any help would be appreciated.
[SUB]RM task_last
MKDIR /mnt/brenda/tmp
Spot request ID: sir-02eapvbw
RMTREE /mnt/brenda/brenda-project.tmp.pre.tmp
MKDIR /mnt/brenda/brenda-project.tmp.pre.tmp
GET http://regono.blendinput.s3.amazonaws.com/FiringRange.blend?Signature=WrbT5$
WRITE[1] 1000000-1999999
WRITE[0] 0-999999
WRITE[2] 2000000-2999999
WRITE[12] 12000000-12999999
WRITE[6] 6000000-6999999
WRITE[5] 5000000-5999999
WRITE[8] 8000000-8999999
WRITE[3] 3000000-3999999
WRITE[10] 10000000-10999999
WRITE[9] 9000000-9999999
WRITE[4] 4000000-4999999
WRITE[7] 7000000-7999999
WRITE[13] 13000000-13667679
WRITE[11] 11000000-11999999
RMTREE /mnt/brenda/brenda-project.tmp
MV /mnt/brenda/brenda-project.tmp.pre.tmp /mnt/brenda/brenda-project.tmp
RMTREE /mnt/brenda/brenda-project.tmp.pre.tmp
PROJ_DIR /mnt/brenda/brenda-project.tmp
Traceback (most recent call last):
File “/usr/local/bin/brenda-node”, line 112, in <module>
main()
File “/usr/local/bin/brenda-node”, line 108, in main
i.start()
File “/usr/local/lib/python2.7/dist-packages/brenda/daemon.py”, line 90, in s$
self.run()
File “/usr/local/lib/python2.7/dist-packages/brenda/daemon.py”, line 143, in $
self.func()
File “/usr/local/bin/brenda-node”, line 105, in <lambda>
func = lambda : node.run_tasks(opts, args, conf)
File “/usr/local/lib/python2.7/dist-packages/brenda/node.py”, line 353, in ru$
error.retry(conf, task_loop)
File “/usr/local/lib/python2.7/dist-packages/brenda/error.py”, line 37, in re$
ret = action()
File “/usr/local/lib/python2.7/dist-packages/brenda/node.py”, line 164, in ta$
task.msg = q.read()
AttributeError: ‘NoneType’ object has no attribute ‘read’[/SUB]
Sorry it’s taken me a while to reply it’s been a crazy busy week. Whenever i get done editing all the lines in settings and hitting enter so the cursor is off screen. Then pressing 3 for home, but It just stays on the same screen. Often times it edits one of the above fields without me seeing it. I’ve found i can close the window and log back in and the setting are kept. Now if I go to 2 from the main menu the screen is as I posted it above. If I press 1 nothing happens I can’t type or navigate back to the home screen.
Hey, I am also sorry for the long response time. I’ve really been swamped recently. As for the issues you’ve been experiencing, I’m not entirely sure what is going on, as I have been able to edit settings fine on my end. It turns out Lumpengnom’s error was not in fact the same.
One thing I noticed from your screenshots is that you are using a client other than Putty to access your instance (something in Google Chrome?). I’ve always done my testing with Putty and that’s what I designed the images for, so before we do anything else to debug this issue I would recommend that you try connecting to your instance through Putty. It sounds dumb but there are sometimes subtle differences between SSH clients in terms of how signals are sent and interpreted. Let me know how it goes.
EDIT: Also, thanks for the tip on starting a VPC. That must be new, as when I made the guide I do not remember having to do that. When I get the time, I’ll do an update to the guide to add that. Glad you figured it out though.
So I think what may be going on is that you have no tasks in your work queue or they are wrong somehow. Are you sure you added render tasks before rendering? Can you see them in the Task Manager screen?
I started out using putty but it had the same errors. Next I tried MobaXterm_v8.1 that’s whats in the screen shot. Today I install cygWin64 and tried that all have the same behavior. My guess is that I have something configured wrong in the VPC idk.
Ok so I didn’t know about the work que. It seems like its working now. I have one question. If I add more instances do they all work together on a single render or do they all need to render a certain frame range of an animation?
If you add more instance each instance will grab a frame and render it. Only one instance will working on a single frame.
If you want more than one instances working on a single frame - if you have a still frame that takes very long to render for example - you can use subframe rendering. You can add subrame tasks in the Manage Queue menu. I haven´t tried this you yet, though.
Thanks for the answers, I have another one. I want to try gpu rendering but I am working on a mac and I am unable to set the render to gpu before sending it to the s3 storage. Can I change the device blender uses to render after i sent it with renderbot? Or is there way to cheat it on a mac to have that option selected before sending it to s3?