cooperative-computing-lab / cctools

The Cooperative Computing Tools (cctools) enable large scale distributed computations to harness hundreds to thousands of machines from clusters, clouds, and grids.
http://ccl.cse.nd.edu
Other
130 stars 111 forks source link

vine: report if the library for a task has not been installed #3789

Open btovar opened 2 months ago

btovar commented 2 months ago

Currently tasks sit in the queue forever if the name of the library they need is not found. At least this should be reported to the debug log with a notice.

dthain commented 1 month ago

Thinking about the possible ways to do this: 1 - Reject the task submission in vine_submit. 2 - Check for library presence in vine_schedule and warn. (We already do this but don't warn.) 3 - Check for library presence in vine_schedule and return the task as a failure.

Is there any case where the function might be submitted before the library?

Jin has already implemented cases where excessive failures of the library cause the removal of the template, and then presumably should cause failure of the waiting functions.