-
Notifications
You must be signed in to change notification settings - Fork 10
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
MDsrv on raspberry pi #48
Comments
Hey Sergio, you can just use MDAnalysis, we used mdtraj as a lighter version but MDAnalysis is now quite powerful and covers mostly all file formats. And yes, we only use it to read the coordinates. Cheers, |
I commented-out a few functions in mdtraj and installed it. I've been able to install MDAnalysis. I will try to see if it works fine and keep you posted. |
Hello,
Can anyone help me? |
By the way, running mdsrv just on the terminal works fine. |
hm, that looks like a numpy error - could you check if that is individually working? (and while you're on it, it's also good, as a backup, to check if MDAnalysis is working or missing some dependencies ;-)) |
On the commandline, mdsrv works fine so MDAnalysis must be fine. Numpy can be imported on the commandline too and I have tried doing some simple calculations and they look fine... from numpy import _distributor_init works fine too. |
just for clarification, you can start the MDsrv from the command line & visualize & run simulations from there but have problems with the deployment? Can you also check if you have different numpy versions installed? might be a dependency problem still... But strange... MDAnalysis can still have some dependency problems which do not arise when you just import it (bug on their side). That's why I recommend trying to import MDAnalysis and e.g. check the version of it. When that works, this is of course not the problem (quite often so far it was the main problem that's why I posted it also here). |
On the commandline I can visualize simulations by calling mdsvr. So it must be the deployment. MDAnalysis seems to be able to load trajectories fine in the commandline. I will workout the numpy versions problem, I am not sure how to check this. I am using a virtual-environment which I thought would take care of this. |
In the environment and on the system I have numpy 1.18.3. Perhaps the problem is that it is using python2 which I have also installed. |
oh yes, python2 can be an issue! MDAnalysis only works with python3, so a mixup could result in not being able to run it. Still strange with having the virtual env. |
It seems that it was using the systems's python binary which is python2. I don't know why it was picking it up. I made some changes and I don't know how, I fixed it. |
okay, that sounds great! I totally forgot to mention, for the deployment, you have to set your virtual environment correctly (in the wsgi file) - maybe there an error occurred too. Of note - how is it running on your raspberry? |
I haven't tried much because I am getting an error when loading a trajectory. This is using mdtraj with commented stuff. I import a .gro file and then onto it I import an .xtc. But when I hit the play button I get this error:
I am not sure if have commented out too many thing of mdtraj (this is the reason I am trying to use mdanalysis which is not hardware accelerated). |
To comment mdtraj out, you can just set
both to I guess the trajectory is fine (MDsrv out of raspberry is running it - or you took one from the example directory) & also MDAnalysis is working from python (e.g. If that's all fine (plz check!), give me your installation protocol/environment, aso, so I can try to reproduce the error on my Raspberry. |
In which file is this? I don't see it in setup.py. |
I get the same problem on the commandline. I only checked a single structure on the commandline, last time. |
I have installed mdsrv with mdanalysis properly and now on the commandline I can import trajectories. But now I have again the problem with numpy in the deployment which I don't understand at all. It seems not to be the python2 thing because I removed the systems python2 binary and the problem remained... |
Since I don't know how I fixed it last time. I don't know how to fix it now. |
okay - can you check |
I am pretty sure it is the deployment that is not setting up the right the environment or picking up things from where it shouldn't. On the commandline it still works fine. In the wsgi I have: The output of /var/log/apache2/error.log is
I have no idea where this python2.7 comes from... I have removed the python2 binaries from /usr/bin. What do you mean by
|
looks like you're using mod_wsgi of python2:
Could you re-install that (check for reference this link)? |
I have done this and now I get this other error. By the way, thanks a lot for your patience and help.
|
ups, there is a Where did you take the wsgi from (wiki or from the mdsrv.wsgi.sample from here https://github.com/arose/mdsrv/blob/master/mdsrv.wsgi.sample)? Need to update that ;-) And sure, I'm happy to help (sorry when replies sometimes take a bit 😅 |
Don't worry you are being very responsive. Now it works! I took it from http://nglviewer.org/mdsrv/installation.html#quickinstall |
Nice! 🎉 Thanks for the reference - will update that file! Let me know if something else comes up - also how it's running for you (I had back then some issues with the resolution but I only have an old raspberry pi). |
Hello,
I am trying to use MDsrv on a raspberry pi server. Since the raspberry pi has ARM architecture it cannot use all of mdtrajs features since a lot of them are hardware accelerated. Does MDsrv use mdtraj extensively or just to load structures, which should work on the pi? In addition, is there a version or some posibility to use MDAnalysis rather than mdtraj which might be more compatible.
Thank you very much for your help,
Sergio
The text was updated successfully, but these errors were encountered: