Maya 'thin client' startup times?

136 views
Skip to first unread message

AK Eric

unread,
Feb 28, 2018, 4:25:47 PM2/28/18
to Python Programming for Autodesk Maya
We're researching using a Maya 'thin client' distro:  Running Maya off a server instead of standalone installs.  I'm tired of finding out someone is 3 versions behind (or never installed the updates / patches) and that's why they're having so many bugs...

The only major problem we've hit is startup times:  Maya 2016 will have the UI launch in around 20 seconds, and it's fully usable in 30 (stops being herky-jerky in the perspective view).

But the thinclient takes 45 sec for the UI to launch, and remains super herky-jerky for another minute:  Basically, unusable for about 1:45 from launch.  

It's even more polarizing in batch mode:  Standalone Maya will boot and be usable in 7 seconds. But the thin client still takes a min thirty.  Ouch.

 IT has been trying to tune the server, but for all I know this is what one should expect... or not: Would appreciate anyone's thoughts on this that has a similar setup.

Justin Israel

unread,
Feb 28, 2018, 5:22:47 PM2/28/18
to python_in...@googlegroups.com
When you talk about a thin client, are you referring to just serving Maya from a network filesystem mount and running it as a local user process? Or are you talking about thin clients where it is actually running on a server and the user is receiving a streamed virtual application?

--
You received this message because you are subscribed to the Google Groups "Python Programming for Autodesk Maya" group.
To unsubscribe from this group and stop receiving emails from it, send an email to python_inside_m...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/python_inside_maya/6f4d0368-6526-449c-b40b-bc2fb14ebc59%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

AK Eric

unread,
Feb 28, 2018, 11:10:05 PM2/28/18
to Python Programming for Autodesk Maya
The first one: "serving Maya from a network filesystem mount and running it as a local user process".

Justin Israel

unread,
Feb 28, 2018, 11:21:22 PM2/28/18
to python_in...@googlegroups.com
On Thu, Mar 1, 2018 at 5:10 PM AK Eric <war...@sbcglobal.net> wrote:
The first one: "serving Maya from a network filesystem mount and running it as a local user process".

Cool. Yea that's what we do at my studio, as well as when I was at Imageworks. I doubt it can be 1-to-1 as fast as loading everything off a local SSD drive; you are subject to network i/o. Bigger studios have the luxury of putting a caching appliance in front of the file server (like Avere) so that hot files can be served out of memory.  I'm not a network/system guy so I don't have a lot of experience with this, but when I was at Siggraph pipeline talks, smaller studios were referring to Pixcache as a cheaper solution.

I am sure your IT people are way more on top of this than anything I could offer. But if strace-ing the load of Maya over a network mount shows that it is spending most of the load time in I/O then maybe the focus should be on caching and optimizing that filer mount for read access.

--
You received this message because you are subscribed to the Google Groups "Python Programming for Autodesk Maya" group.
To unsubscribe from this group and stop receiving emails from it, send an email to python_inside_m...@googlegroups.com.

Luke Harris

unread,
Mar 1, 2018, 3:46:30 AM3/1/18
to Python Programming for Autodesk Maya
Are people using configuration management tools like puppet / chef / salt / ansible / pdq deploy / sccm to make sure workstations have the correct versions of everything? I have no numbers to back this up - but I imagine a network app server will never be as snappy as a local SSD for big apps like maya
Reply all
Reply to author
Forward
0 new messages