Running multiple GPT instances simultaneously by different users on linux machine

Hello step community,

I currently observed that running GPT from different users on a Linux machine does somehow create problems. When a single GPT process is running and another user starts a GPT command, often the error message:

However, as soon as the process is finished, another GPT can be started without issues.

Is it possible to alter some config files etc. to prevent this error from happening?

Thanks a lot

Did you install SNAP separately for each user or are the users sharing a common install? In our lab each user has their own installation and we have not encountered conflicts such as you have, only the expected resource limitations of total RAM in the machine.

@gnwiii Thanks for your reply.

Indeed we are sharing an installation. I already thought that this might be a problem. However, I assume that it must be possible to share an Installation.

We still did not find a solution and I am happy for any help. cheers

I guess this is only happening when you use Sentinel-2 data.
There is a check which leads to this error.

@obarrilero Can you check why this happens and if it can be avoided?

@c3po SNAP and gpt have not really be developed with the idea in mind to have multiple instances running at the same time. So it might work or it might not. For the future, we have ideas to better separate the instances. But this doesn’ help you now.
It could help you if you specify a different user directory for each user/instance.
You can add to the gpt call -Dsnap.userdir=Path/to/user_dir
Hope this helps.

Perhaps it is a write permissions issue in .snap/var/cache/s2tbx for the other users?

Thanks @marpet for your suggestion. When including the -Dsnap command, this error occurs: Invalid OpenJpeg executables, which was also discussed here

So I assume there is still some kind of permission problem.