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
{{ message }}
This repository has been archived by the owner on Jan 6, 2020. It is now read-only.
When using Drive in Linux, after unmounting and destroying the FuseDriveInUserSpace object there is still a reference to the ChunkStore object kept alive which was passed in via shared pointer. This happens even when Unmount() and WaitUntilUnMounted() are called after Mount() returned.
If there are still processes running in Drive or Encrypt, ideally WaitUntilUnMounted() would wait until all of them are finished (or timed out) so no objects remain in memory which could have a pointer to the ChunkStore and other shared resources. This should happen as a guarantee that once Drive is closed no more changes will be made to the ChunkStore and that its state at that point can be considered final.
This happens using the latest commit of the 'next' branch and must have been introduced during the last 3 weeks. The problem can be observed when running SigmoidPro, there is a lack of debug output normally printed by ~RemoteChunkStore().
The text was updated successfully, but these errors were encountered:
Sign up for freeto subscribe to this conversation on GitHub.
Already have an account?
Sign in.
When using Drive in Linux, after unmounting and destroying the FuseDriveInUserSpace object there is still a reference to the ChunkStore object kept alive which was passed in via shared pointer. This happens even when Unmount() and WaitUntilUnMounted() are called after Mount() returned.
If there are still processes running in Drive or Encrypt, ideally WaitUntilUnMounted() would wait until all of them are finished (or timed out) so no objects remain in memory which could have a pointer to the ChunkStore and other shared resources. This should happen as a guarantee that once Drive is closed no more changes will be made to the ChunkStore and that its state at that point can be considered final.
This happens using the latest commit of the 'next' branch and must have been introduced during the last 3 weeks. The problem can be observed when running SigmoidPro, there is a lack of debug output normally printed by ~RemoteChunkStore().
The text was updated successfully, but these errors were encountered: