I purchased WebODM for Mac today and downloaded the files.
When I run WebODM manager it says it is downloading Docker and then launching docker. Then I get a message saying Docker is corrupt and should be moved to trash. That’s as far as it gets. After a while I get a message saying it was having trouble running docket and to start it manually, but Docker does not show in the application launcher.
System is a Mac Mini M1 8GB running MacOS Ventura Version 13.1
Any assistance would be appreciated. I’d prefer to fix this instead of getting a refund and having to install Docker and all that manually.
If you have not already, please reach out to [email protected] for order/installation support.
Two things are at play here, most likely (from what we’re seeing roll in now as Ventura reaches more machines:
MacOS GateKeeper is even more aggressive than prior releases, and uses misleading verbiage about applications not from/signed by the Apple AppStore (it will say they are corrupt/damaged, which is not true).
Safely open apps on your Mac - Apple Support
Docker versions above v4.12 seem to have issues on some MacOS machines (usually M1 Max machines, but not always/only):
Docker Desktop release notes | Docker Documentation
Some steps to try:
- Uninstall current Docker installation
- Download Docker v4.12 and move installer file out of Downloads to Desktop
- Read/Follow directions from Apple about running Apps from outside the AppStore
- Install WebODM Manager (if not already installed and allow-listed per Apple documentation)
- Install Docker v4.12 per Apple documentation
- Launch WebODM Manager
- Wait for Docker to launch
- Press “Update” on WebODM Manager and wait for it to update/launch WebODM
Thanks for the response.
I was able to get this to work after uninstalling everything, installing docker using the Apple Silicone version found at the direct URL (Install on Mac | Docker Documentation) and then installing the WebODM Manager.
At that point it seemed to recognize that Docker was already installed and everything progressed as you would expect.
Is it possible that the version of Docker that is being downloaded by the WebODM manager is the Intel version? If so, perhaps you can update that to download the correct version based on the machine that is being run.
Now…that said. I have not managed to get it to actually create a map yet. I’ll try a few things and, if I can’t find a solution, will create a separate post on that.
We download the correct version for the machine’s architecture, the problem is we fetch the latest and on many machines, anything above v4.12 seems to be broken for some reason.
You’ll likely need to tune your Docker Resource Allocation:
The default setup is very conservative, so you should have some room to expand RAM (leave no less than 8GB for MacOS), CPU count (can safely take all), swap (push it to full), and disk (256GB is a safe minimum for most datasets, though you may need more [though you not have enough space to go larger]).
At the moment I’m running on a Mac Mini with 8GB of ram.
While that doesn’t sound like much, the unified memory setup means it has no problem with tasks like editing a 4k timeline in Davinci Resolve. However, it does seem that WebODM needs actual ram to work.
I have a 16GB of the M2 mini on order so will test once that one comes in by allocating more to Docker. At the moment though I can’t get it to do anything on the 8GB unit. I’ll let you know how it goes once the new setup comes i.
I have a MacStudio M1 and have been wrestling with Docker for just about 2 months- first let me say I know ABSOLUTELY NOTHING about this!
Secondly, IMHO- your setup sounds like it might be a bit light for webODM. Mapping needs more resources than photo and video.
Definitely contact support at that email address they provided.
Thirdly- what was that “Docker Silicone” thing you spoke of?
I’ve been chatting with support (very responsive) but we moved the conversation here so that others can benefit from it, even if at my expense.
It seems if you manually install Docker (easy to do with the link I provided earlier) then the rest of the install goes as expected and everything works.
The “Apple Silicone” (not Docker Silicone) refers to the fact that Apple are now making their own CPUs.
Personally, I was never an Apple guy but their new CPUs are amazingly fast and less money than what I would have to spend to get comparable performance using a PC, so I have switched over for the time being.
The new Apple system uses “unified memory” which is to say the CPU and GPU share memory, and it also uses the SSD swap much more effectively. This means that you can get away with much less memory than you would need on (say) an Intel chip.
Other resource heavy programs such as Davinci Resolve have been optimized to work with this and so less ram is needed. It seems that Docker/WebODM has not and still needs regular ram.
I guess that leaves me with a few choices.
- Get more ram - I have a Mac with more ram on the way and I will be returning this one (long story). It will still only be 16GB though so we will see if that is enough for what I need
- Keep my old and failing PC which has 32GB of ram around and use it only for handling maps. I’ll do this for a while anyway
- Process in WebODM lightening - I have quite a few credits from referring people through my training videos like this one Dronelink 3D Modeling - End to End - YouTube
- Install in the cloud on something like AWS
I might need to do a mix of these in fact - depending on need. The nice thing is, I don’t really have to choose one over the other. They all work.
We will see how things play out once the new machine comes in later this week.
Wonderful! That is exactly what we want for folks, the ability, and freedom, to work how it suits you best!
A quick update. I received the M2 Mini today with 16GB of ram.
I allocated most of that to Docker and ran a test that previously failed on the M1 8GB version and it all processed with no issues. It also gave all the resources back when I closed Docker, which is nice.
It will be interesting to see how it handles some larger projects but, for now, things seem to be working well.