cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Brandon Williams (JIRA)" <j...@apache.org>
Subject [jira] [Created] (CASSANDRA-12217) Assassinate should use the current timestamp for the generation instead of incrementing
Date Fri, 15 Jul 2016 20:34:20 GMT
Brandon Williams created CASSANDRA-12217:
--------------------------------------------

             Summary: Assassinate should use the current timestamp for the generation instead
of incrementing
                 Key: CASSANDRA-12217
                 URL: https://issues.apache.org/jira/browse/CASSANDRA-12217
             Project: Cassandra
          Issue Type: Bug
            Reporter: Brandon Williams
            Assignee: Brandon Williams
             Fix For: 2.1.16


In CASSANDRA-8113 we made gossip ignore generations more than a year in the future.  While
that is good, what can happen is this: you errantly have a machine start with a clock at,
say, epoch+1, making the generation 1.  Now, you're kind of stuck, because all other nodes
will ignore you until you get up to now - 1 year, which from 1970 will take a lot of assassinates
since it only increments the generation.  Instead, I propose we use the assassinating node's
timestamp as the generation, so that we have a way out of this aside from discarding the IP
address for 3 days.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message