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
Currently we have a "delete data in repo/tmp older than a day" policy. This is suboptimal for many reasons.
A workflow we want to support is doing background pulls with explicit deployments. In the case of interrupted pulls, we don't want to GC staged objects for the "latest pending commit", even if a day elapsed between pulls. However, we do want to GC staged objects not referenced by either the "new commit" or any of our current deployments.
One way to implement this might be for ostree_repo_pull() to actually do the write objects (but not commit/update the ref) if the failure was due to HTTP (but not anything else, like ENOSPC). This would move the objects out of staging, so we wouldn't have to worry about the one-day cleanup policy.
This would still mean though that ostree admin cleanup would prune those objects...we could consider writing a special "partial" ref for pending updates? Or, maybe that's a feature, and if you don't want those objects to be pruned, don't call cleanup?
The text was updated successfully, but these errors were encountered:
Currently we have a "delete data in repo/tmp older than a day" policy. This is suboptimal for many reasons.
A workflow we want to support is doing background pulls with explicit deployments. In the case of interrupted pulls, we don't want to GC staged objects for the "latest pending commit", even if a day elapsed between pulls. However, we do want to GC staged objects not referenced by either the "new commit" or any of our current deployments.
One way to implement this might be for
ostree_repo_pull()
to actually do the write objects (but not commit/update the ref) if the failure was due to HTTP (but not anything else, likeENOSPC
). This would move the objects out of staging, so we wouldn't have to worry about the one-day cleanup policy.This would still mean though that
ostree admin cleanup
would prune those objects...we could consider writing a special "partial" ref for pending updates? Or, maybe that's a feature, and if you don't want those objects to be pruned, don't call cleanup?The text was updated successfully, but these errors were encountered: