-------- Forwarded Message --------

Subject:	Re: [Chimera-users] run chimera remotely over nice dcv
Date:	Sat, 21 Apr 2018 13:20:11 +0200
From:	Mihajlo Vanevic <m.vanevic@uu.nl>
To:	Greg Couch <gregc@cgl.ucsf.edu>

Dear Greg,

Thank you very much for the prompt reply.

About ssh -X, indeed putting the compatible libGL.so from local machine
to Chimera's lib folder does the trick.

About Nice DCV, I have contacted Nice DCV developers and will
let you know if the problem is solved.

With best wishes,
Mihajlo

   

Hi Mihajlo,

We don't have any experience with Nice DCV, perhaps some other user on the mailing list can help.  But guessing at possible problems, make sure that Chimera is displaying locally in Nice DCV virtual machine (the DISPLAY environment variable should be ":0"), and the remote display is done using the Nice DCV client.

If, instead, you're using X11's remote display capability, then you wouldn't need to use Nice DCV, any virtual machine technology would do (or not do) -- the trick there is to have compatible OpenGL drivers on both the virtual machine and your local computer (running the X11 server).  Assuming compatible versions of Linux on both ends, you could copy libGL.so from your local machine to the Chimera's lib directory on the remote machine, and it should work.

    Good luck,

    Greg

On 4/20/2018 5:07 AM, Mihajlo Vanevic wrote:

Dear Chimera users and developers,

We are trying to run Chimera remotely on our cluster via Nice DCV
visualization software:

https://www.nice-software.com/products/dcv

This package provides GPU acceleration for OpenGL and
DirectX applications. It indeed works for most of apps, however with Chimera
we get the following startup error

X Error of failed request:  GLXBadContext
  Major opcode of failed request:  152 (GLX)
  Minor opcode of failed request:  5 (X_GLXMakeCurrent)
  Serial number of failed request:  73
  Current serial number in output stream:  73

Our cluster visualization node has nvidia GeForce GTX 1080 Ti cards
and nvidia drivers 384.81

Chimera works if we use software mesa rendering, but it would be great if
we can make it work remotely with hardware rendering.

With kind regards,
Mihajlo Vanevic
Cryo-EM group, Utrecht University