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
We should keep more epochs on the master. Right now a node that is catching up fro very far behind if it loses connection and tries to reconnect will run into an issue of too many epochs. Increasing the number of epochs recognized by the master will help with this situation though not totally resolve it. Another option is in this case to start over in the pre-replica node that is catching up (though this can obviously lead to problems)
The text was updated successfully, but these errors were encountered:
We should keep more epochs on the master. Right now a node that is catching up fro very far behind if it loses connection and tries to reconnect will run into an issue of too many epochs. Increasing the number of epochs recognized by the master will help with this situation though not totally resolve it. Another option is in this case to start over in the pre-replica node that is catching up (though this can obviously lead to problems)
The text was updated successfully, but these errors were encountered: