You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We recently had a "bring your own workflow" workshop for our GA4GH Cloud stack, and preparing for that, we were reminded that one can't just reliably run any workflow in the cloud via WES/TES. Dependencies need to be embedded in a workflow, unambiguously and for each step, either as a container image or a Conda package. Of course, having a fully "containerized" or "condarized" workflow still does not guarantee that the workflow will run on any WES or WES/TES stack - but it would arguably be a bare minimum requirement.
Currently, there is no way of knowing whether a workflow resource on TRS meets this requirement - one would have to run the workflow and inspect the error stack.
I don't think that evaluating whether a workflow has all dependencies embedded in this way is a job for TRS (probably similar to #224 in that respect). However, I still wanted to raise this issue to spark a discussion on how this might possibly be achieved. Perhaps this could feed into the discussion on #223 - which would allow users to report back usage stats/info. Depending on how such a solution might be designed, workflows and tools could be amended with important info on workflow validity, cloud compatibility etc.
┆Issue is synchronized with this Jira Story
┆Project Name: Zzz-ARCHIVE GA4GH tool-registry-service
┆Issue Number: TRS-61
The text was updated successfully, but these errors were encountered:
We recently had a "bring your own workflow" workshop for our GA4GH Cloud stack, and preparing for that, we were reminded that one can't just reliably run any workflow in the cloud via WES/TES. Dependencies need to be embedded in a workflow, unambiguously and for each step, either as a container image or a Conda package. Of course, having a fully "containerized" or "condarized" workflow still does not guarantee that the workflow will run on any WES or WES/TES stack - but it would arguably be a bare minimum requirement.
Currently, there is no way of knowing whether a workflow resource on TRS meets this requirement - one would have to run the workflow and inspect the error stack.
I don't think that evaluating whether a workflow has all dependencies embedded in this way is a job for TRS (probably similar to #224 in that respect). However, I still wanted to raise this issue to spark a discussion on how this might possibly be achieved. Perhaps this could feed into the discussion on #223 - which would allow users to report back usage stats/info. Depending on how such a solution might be designed, workflows and tools could be amended with important info on workflow validity, cloud compatibility etc.
┆Issue is synchronized with this Jira Story
┆Project Name: Zzz-ARCHIVE GA4GH tool-registry-service
┆Issue Number: TRS-61
The text was updated successfully, but these errors were encountered: