Generative design stuck on 5%

[Anonymous]

I have been experimenting a lot with Generative design lately, as it is used in a class I'm in. 

However, since yesterday all my designs get stuck after they are sent to the cloud. In the job status menu the solve is stuck on 5% and in the explore menu no outcomes appear, even after 12+ hours. It appeared to start after I cancelled a study that seemed to not converge. It's almost as if that study "clogged" the system in some way. 

I couldn't find any solution to this while searching the forums, except for one other post that quite closely resembles my issue: "generative design is not generating solutions" . However, that post sadly lacked a solution. 

Grateful for any help - I just hope I can get it working again before my delivery deadline on Monday 

[Ben-Weiss · Accepted solution]

Hi Jacob,

Thanks for reaching out. It is unusual for it to take this long to be able to see outcomes in Explore (though it is normal for the progress to be stuck at 5% until the generative job ends). I looked at our system's logs, and I see

 - Two jobs started this morning (6 am Pacific time; 15 hours ago) which are still running

 - A dozen or so jobs started about 8 hours later which show as cancelled

 - Two jobs started shortly thereafter which show as "Pending..." (waiting in line to start)

The two old jobs that are still running are caught in a glitch in our system. I have notified the relevant team, but most everyone is gone for the weekend now, so it may not get looked at until Monday. The good news is, they are not common glitches, and making any small change to your preserve geometry should keep them from reappearing. If you continue to have problems, let me know and we'll investigate further.

I don't remember, but there may be a limit to the number of education jobs each user can have running at a time, in which case your new submissions could be waiting for your old submissions to finish so they can start. There should be a 24-hour timeout on all jobs, so hopefully the stuck jobs will time out and the new submissions will start running tomorrow. Unfortunately they don't give me a "cancel" button to kill your jobs manually and unclog things.

Sorry not to be more helpful!

Ben

[Anonymous]

Hi again,

I cancelled all studies that were in the queue, and after a while I was able to start a new generative study and actually get some outcomes to show up. Sadly, it seems that the same problem occurred again. The outcomes have gone through a reasonable amount of iterations and look completed upon inspection, but are stuck as "processing" in the explore menu. This has been the case for 12 hours. I tried changing the preserve geometry like you suggested, but it seemed not to have an effect. 

I have ran several similar studies as I am exploring the impact of starting shape on the outcomes, and it the preserve geometry has been the same for a number of studies. However, the first study that froze might have been the first time I tried running two studies with the same loads and preserve, but different starting shape simultaneously.

Thank you so much for your help - I'm doing these gen designs for a school exam that was supposed to be delivered yesterday, but luckily I got an extension because of this issue.

[Ben-Weiss]

I see two jobs from yesterday that are stuck in the same glitch as before; sorry my suggestion didn't fix the problem entirely. The team responsible for that piece of the system would like to investigate, and would appreciate it if you can share your setup with us by exporting an f3d (File->Export->Fusion 360 Archive) so we can dig into the issue. If you would rather dm it to me than post your final exam solution publicly, that would be fine.

Again, sorry the system failed you!

Ben

[Anonymous]

No worries, I got an extension on the paper, so everything should be fine as long as I get it working again within the week.
The generative design is just an exploration using the GE bracket challenge from grabCAD, so sharing it here should be no problem.

Again, thank you for your efforts!

[Anonymous]

I have been able to run through a couple of studies with the same setup now, but it seems that I can not create new designs from the outcomes generated. Is this a known issue?

[Ben-Weiss]

Hi Jacob,

This may be related to the issue we found with outcomes taking forever to generate for your model. The team is investigating a fix, but it won't be ready for a while yet.

If you don't get an "outcome preview" when looking at the last iteration of the outcome you want to export, it's a pretty good chance that exporting is going to fail (and it's probably caused by the issue we're looking into). I will check with the team to see if they have a workaround that could help you get outcomes to export.

Thanks for persevering!

Ben

[Ben-Weiss]

Unfortunately, the team responsible for the component that's failing hasn't had time to investigate the issue. I will keep tabs on the issue and let you know if/when they have any suggestions.

Sorry I'm not being more helpful here!

Ben

[martin_marinov77]

Hi Jacob,

Thanks to Ben who alerted me to the urgency of your issue, I had a look at your setup and identified the cause of the export failures.

In the preserves there is a specific type of geometry at the top of these rolling ball blends, known as "knife" edges, that is not supported by the GD conversion:

The reason knife edges are not supported is that the model thickness converges to 0 as you approach them and therefore there is no physical way of producing such shapes. We should be alerting the user when they supply such input to GD solves, but we have not implemented this yet.

To resolve this problem you have to setup the preserve geometry differently and avoid knife edge geometry. Usually, when I've seen this model in the past, it had the following preserve geometry: 

The additional material avoids the knife geometry. If this is too much material for your setup, you can try cutting off the part above the blend. Alternatively,  you can either blend or chamfer the knife edges.

Hope this helps and please let us know if you still experience issues after fixing this problem.

Thanks,

Martin

[DrHavelock]

I am having this same issue, the Generative Design is stuck on 5% and every study Fails. I have tried setting up the study several different ways with structural loads, constraints and preserve geometry. Finally, I opened a previous study that I have successfully run and obtained a successful result  - when I run that study again it Fails.

Can someone please help?

Thanks, Slade

[Ben-Weiss]

Hi Slade,

I looked at the logs for a few of the studies you ran on Sunday and Monday and saw mostly errors on our side. We've been dealing with an outage over the last few days that has caused some jobs to fail. One root cause has been identified and fixed; we're paying close attention to see if we have completely resolved the issues.

Good news is I don't see any errors that suggest the failures are caused by something incorrect with your problem setup.

Can you try generating a clone of one of your studies which previously succeeded and see if it works now?

Thanks,

Ben

[Anonymous]

I am facing a similar issue. If I try to run my generative design study, every outcome stops generating after 2 or 3 iterations. I have tried to setup my study in various ways, but the outcomes always "fail" after just a few iterations. Other studys, which previously just worked fine bring up the same issue. 

Maybe there is some way you can help me.

Thanks, Maxi

[Ben-Weiss]

Hi @Anonymous,

Thanks for reaching out. I checked a few of the models you've run over the last 24 hours and I do not see anything that is caused by an issue on our side. Since this seems to be a different issue, please start a new thread and include some additional information about the setup you are using and the problems you are seeing.

Thanks,

Ben

[Anonymous]

I'm also facing a similar issue, everything just hangs on 5% then fails after an hour or two. I've tried to solve the example files supplied in fusion 360 but I get the same results. Any suggestions would be greatly appreciated.

Regards

Nick

[gbalhoff5NAJ2]

I am having the same issue. I have tried about 5 times to run a simulation and I don't think it's processed through yet. I let one run overnight and one all day. Admittedly the first iterations were probably not ideal setups so I made some tweaks. But now it's about as simple as it gets and it has been running for hours still stuck. @Ben-Weiss 

[Ben-Weiss]

Hi @gbalhoff5NAJ2,

Thanks for posting! So we can best address your issue and to make it easy for others with similar problems to find answers, could you please start a new thread and include some pictures, or if possible an f3d file (File->Export->Autodesk Fusion 360 Archive) of the simulation that's giving you trouble?

Thanks,

Ben

[celso.messias]

I'm with the same issue. I had tried clone study, save as new file, change loads and nothing.

[Ben-Weiss]

Hi @celso.messias,

Thanks for reaching out! So that we can best address your issue and so that others can easily find solutions helpful to them, please start a new thread on the forum.

Thanks,

Ben

[kevin_robertEQT5M]

Hello, 

i seems like I have the same trouble too. Could you check for me too please ? 

[Ben-Weiss]

Hi @kevin_robertEQT5M,

They changed the way our back-end systems work and it's now much harder for me to track users' jobs from my side, since I'm not a support person. If you've been waiting more than 24 hours, start a new thread to increase visibility among the people best equipped to deal with this issue.

Thanks and sorry not to be more helpful,

Ben