Commits on Aug 28, 2019

1. [FIXED] Clustering: Snapshot skip sequence on error …

   If the leader had an issue sending a requested message, it would
   simply log the error but continue with the next. The consequence
   would be that the follower would get a "gap" and empty the store
   up to that sequence assuming this was the new first available
   message.
   The missing `return` statement has been added, but there would
   still be a situation where the follower could wrongly assume that
   the mismatch in sequence between the request and the response
   meant that this was the new first available message in the leader.
   The leader is now sending - through the reply subject - information
   that indicates that this is the first message. If not, the follower
   has to verify that when getting a mismatch sequence, if this is
   because messages were lost or because this is indeed the first
   available. It is possible through a request of a single message
   as opposed to a batch.
   
   Also replaced some panic() calls due to Unmarshal() where a returned
   error is more appropriate.
   
   Resolves #921
   
   Signed-off-by: Ivan Kozlovic <ivan@synadia.com>

   

   kozlovic committed Aug 28, 2019
   Configuration menu

   • View commit details
   • Copy full SHA for 5c6d507
   • Browse repository at this point

   Copy the full SHA

   5c6d507 View commit details

   Browse the repository at this point in the history