Skip to content
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

computational backend shall make use of multipart links for external clusters #3217

Open
Tracked by #617
sanderegg opened this issue Jul 5, 2022 · 3 comments
Open
Tracked by #617
Assignees
Labels
a:dask-service Any of the dask services: dask-scheduler/sidecar or worker a:director-v2 issue related with the director-v2 service t:maintenance Some planned maintenance work

Comments

@sanderegg
Copy link
Member

currently external clusters are limited to 1 single presigned link, thus limited to 5Gb uploads

@sanderegg sanderegg self-assigned this Jul 5, 2022
@sanderegg
Copy link
Member Author

@elisabettai
currently the ENV VAR STORAGE_DEFAULT_PRESIGNED_LINK_EXPIRATION_SECONDS is set as 6 hours until the computational sidecar can properly request an upload link

@sanderegg sanderegg transferred this issue from ITISFoundation/osparc-issues Jul 25, 2022
@sanderegg
Copy link
Member Author

sanderegg commented Sep 9, 2022

webserver

  • start computation
  • creates a token, pass it to director-v2 (ideally with expiration time)
  • delete at the end of computation

director-v2

  • director-v2 creates a token to access the osparc-api for the project or node that must run (this token is then deleted after the project/node has run) for output files and log file
  • instead of passing the links to the output files, the director-v2 passes the token to the dask-sidecar and the url of the public api

dask-sidecar

  • after the computational service has run, the sidecar will use the osparc public-api to upload any retrieved output file
  • same will happen with the log file

@elisabettai
Copy link
Contributor

elisabettai commented Feb 1, 2023

fyi, @sanderegg we have now a comp. service (tvb-recon) that takes potentially more than 6h to run. This is expected according to TN.
I'm still testing if it all works on dalco staging.
As a precautionary measure I've set STORAGE_DEFAULT_PRESIGNED_LINK_EXPIRATION_SECONDS in the webserver (on dalco staging) to 43200. Maybe we should change the default value to accommodate this use-case.

@sanderegg sanderegg added t:maintenance Some planned maintenance work a:director-v2 issue related with the director-v2 service a:dask-service Any of the dask services: dask-scheduler/sidecar or worker labels Aug 20, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
a:dask-service Any of the dask services: dask-scheduler/sidecar or worker a:director-v2 issue related with the director-v2 service t:maintenance Some planned maintenance work
Projects
None yet
Development

No branches or pull requests

3 participants