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
Is your feature request related to a problem? Please describe
Given that remote store garbage collectors are aware of pinned timestamps, it becomes important to communicate any change to pinned timestamp to all the nodes in cluster within pre-defined SLA.
Describe the solution you'd like
To make sure each node gets the updated pinned timestamp details within X mins, in this solution, we run a scheduler on each node with period of X/2 mins (this is to make sure we run the scheduler at least twice).
This scheduler will fetch the list of pinned timestamps and store it along with timestamp of the successful fetch.
The timestamp of successful is required for garbage collector to identify if it is accessing stale pinned timestamp list and in such cases, garbage collector needs to be skipped.
Related component
Storage:Remote
The text was updated successfully, but these errors were encountered:
Is your feature request related to a problem? Please describe
Given that remote store garbage collectors are aware of pinned timestamps, it becomes important to communicate any change to pinned timestamp to all the nodes in cluster within pre-defined SLA.
Describe the solution you'd like
Related component
Storage:Remote
The text was updated successfully, but these errors were encountered: