Hi All,

As per usual I have left my uni project to the last minute, and was banking on using the HQueue Cloud service to get me over the line and… no dice.

Every time I try to roll the cloud render service the HQueue dialogue just hangs on the ‘Launching a computer in the cloud’ page. Have checked on my Amazon account and no instances are running. Tried on two different machines (OSX 10.6.8 and Windows 7) both running production build 12.0.634 with Apprentice HD licencing.

If anyone has a solution to this issue please help, pretty much you will be my hero.

FYI this issue has been raised recently by dgc289 on the amazon forums: https://forums.aws.amazon.com/thread.jspa?messageID=348890&#348890 [forums.aws.amazon.com]

Sorry if this is a duplicate topic - I cannot find his original post in this forum

*BUMP*

Just going to bump this post up again to see if anyone has a solution.

I have the same problem with hqueue cloud render.
When i launch the console the following line looks wrong to me.

hython: BUG in libdispatch client: Do not close random Unix descriptors

some suggestions ?

thx

jopa79
I have the same problem with hqueue cloud render.
When i launch the console the following line looks wrong to me.

hython: BUG in libdispatch client: Do not close random Unix descriptors

some suggestions ?

thx

Hello,

So what happens after you proceed with your cloud submission? Does a web page at least open up that says, “Launching a computer in the cloud”?

Cheers,
Rob

Hi Rob,

no, nothing happens.

thx

What OS are you on?

Also, are you able to run hython from the terminal?

–Rob

Hi Rob,

i'm running H12 1.125 on OS X 10.8.2. I can run hython from the houdini shell without getting errors. I checked all requirements, following the guide, file dependencies are all green but after hit “yes, procced” nothing happens.
cloudsubmit.log says “cloudsubmit starting up” but no web page coming up.

Ah. I had another user recently run into troubles with submitting cloud jobs from OSX >= 10.7. The problem had to do with a bug in Python >= 2.6.4 on OSX.

Try starting Houdini like so from the Houdini Shell.terminal app:

no_proxy=“ ” houdini

That temporarily sets $no_proxy to a non-empty value (i.e. a single whitespace character) and then starts Houdini.

See if this workaround helps. Otherwise, I'll probably have to send you a patch file that will help us track down the issue on your machine.

Let me know how that goes.

Cheers,
Rob

Hi Rob,

thanks for your help, now i'm one step closer :wink: i think.
I start houdini with your workaround and my browser appears.
After more than 20 minutes of waiting nothing happens (Launching a computer in the cloud).
The EC2 Instance would not start (my scene file is just 1,5M). I click on the amazon console link and the EC2 Dashboard tells me that i have 0 running instances.

thanks…

Awesome. We're getting there.

If you see the web page, then that means that Houdini was able to get to a point where it spawned a daemon process called ‘cloudsubmit’ which is responsible for starting EC2 instances and managing render jobs on the cloud.

So if the EC2 dashboard reports 0 instances, then the cloudsubmit daemon must have hit a glitch when trying to start the instances. This can happen for a variety of reasons such as a network firewall configuration setting or if the EC2 Amazon account that you are using is not subscribed to the Houdini Cloud Rendering service.

To find out what the problem is, we need to examine cloudsubmit's log file. On Mac OSX, the log file is stored in a temporary directory somewhere under /var/folders.

Open a Terminal.app, and run this command to find the log file:

find /var/folders -name “cloudsubmit.log”


Once you locate the log file, you can post it here or send it to me by email and I can help you from there.

Cheers,
Rob

Hi Rob,

…and this was my fault
“..EC2 Amazon account that you are using is not subscribed to the Houdini Cloud Rendering service..” :roll:

Now it works.

I'm guessing that python bug needs to be fixed by apple ?

thank you again.

Glad to hear it's working!

I no longer have the link but if I remember correctly, the bug was logged with the Python dev team.

Cheers,
Rob

hello there

i am also trying to use ec2 cloud computing and i also get stuck at the web page.
i am on win 7..
where do i find the log file on win7…
and also how do i subscribe my amazon ec2 account to Houdini Cloud Rendering service.

thanks your help will be really apreciated

i found my log file with the following eror

“AuthFailure: Subscription to ProductCode 5D6864EE required.”

my problem is to understand how to subscribe to an AMI or whatever it is called on amazon, after 2 hours it seems i am still unable to do so.
plz help

thank you

Hello,

Have you followed the HQueue Cloud user guide? http://www.sidefx.com/images/stories/news/HQueue_Cloud/hqueue_cloud_guide_09nov12.pdf [sidefx.com]

Page 4 of the PDF talks about how to register for the cloud service.

Cheers,
Rob

yes i followed the steps exactly…
every thing is fine but then i get that message in my log file.

"Please visit https://aws-portal.amazon.com/gp/aws/user/subscription/index.html?offeringCode=59D3AAF6 [aws-portal.amazon.com]
2012-12-22 03:03:50,473 ERROR: AuthFailure: Subscription to ProductCode 5D6864EE required."

i can go and lauch any instance of ubuntu or whatever, but i am not allowed to start manually a houdini instance, neither can i from houdini itself
i am using aprentice HD.

since i can lauch any other instance , i believe it isnt related to billing issues.

i have my shared key and secret etc, but it just wont work

what a fail…after posting my last comment i saw the obvious “please check amazon blabla”, it took me further into the registration..
i ll try and see if it works

it is working but now i have no download progress

So it is rendering and downloading
but i ma quiet disappointed with the render time

i launched a simple scenes that takes 3 min for 100 frames on my
12 core machine (dual xeon 3.07ghz 48 mb ram)

hqueue with clouds tells me 45 min..with that setup:

10 machines 64 bit
min client per frame 10
max client per frame 10
frame per job 1
cpus per job 8.

anything i am doing wrong?

i apologise for beeing a bit pushy but i litterally lost all my night to figure out to make that work and the result is not what i expected.

anything i am doing wrong?

In a way, you are.
Small jobs are not suitable for the Hcloud. amazon does not have a “persistent” renderfarm setup. When you send a job, they allocate arbitrary machines and roll out the software first. Then ifds are being generated and then the rendering is started. If you have for example a rendertime of 10 minutes per frame, then you will profit from it. For short rendertimes per frame the overhead is far too great.

You also have to be aware, that the EC2-machines are rather mediocre (even the XL-ones) compared to your rig. You will have to take that into account when calculating the frame rendertime in the first place.