Closed jluethi closed 12 months ago
The coverage rate went from 89.91%
to 89.84%
:arrow_down:
The branch rate is 83%
.
75%
of new lines are covered.
These tasks are not tested in example 03 in fractal-demos and it seems to run well. In 5 runs, I could not reproduce this race condition anymore.
Addresses part of https://github.com/fractal-analytics-platform/fractal-tasks-core/issues/516 by making it less likely to happen.
I'm creating a PR to make this race condition way less likely:
Instead of removing the old group first (slow call), then renaming the new group to the old name, I'll now do the following: Rename the old group to group_name_tmp (e.g. (0_tmp). Then rename the new group to the old group name (e.g. from 0_registered to 0). Both those renamings are very fast. Only after that, remove the tmp folder. The opening of the old zarr group gets a try except to catch the zarr.errors.GroupNotFoundError and would try again 5 seconds later. This should make it very unlikely to hit the issue. It's not closing the issue yet, see description there.
Checklist before merging
CHANGELOG.md
=> we're only adding the task with this PR