galaxycomputationalchemistry / galaxy-tools-compchem

:mega: Galaxy Tools for Computational Chemistry
Apache License 2.0
14 stars 16 forks source link

Tool fails for large dcd trajectories #15

Closed chrisbarnettster closed 5 years ago

chrisbarnettster commented 5 years ago

Received this error from a user running on Galaxy EU.

However the Galaxy Chemical tool box could not process the dcd file. "Error: cannot allocate vector of size 20.5 Gb Execution halted Fatal error: Exit code 1 ().

Following up but probably true for mdanalysis and bio3d. For larger files, either a very large memory system is needed or a chunking strategy must be used.

chrisbarnettster commented 5 years ago

In the interim a suggestion is to strip down the trajectory using catdcd. Bio3d has an issue with allocating large arrays.

bgruening commented 5 years ago

Chris you just need to let me know how much every tool needs. We allocate for every tool special memory. We have enough big memory nodes available so this should not be a problem if we give this tool more memory. Can you send me a list with approximate men usage per tool?

tsenapathi commented 5 years ago

This issue is only for Bio3D tools. Maximum for now 64GB.

chrisbarnettster commented 5 years ago

Thanks ! Yes as Tharindu mentioned it is only the Bio3D suite for now. This would depend on the simulation size. I think a 64Gb threshold is fair.

bgruening commented 5 years ago

Hi! See here: https://github.com/usegalaxy-eu/infrastructure-playbook/commit/65eaab0d88792611d6f07d7909b2461bcc58adeb

Should be fixed.