[GitHub] tysonnorris commented on issue #2531: Share bookkeeping data across controllers

[gi...@git.apache.org]

tysonnorris commented on issue #2531: Share bookkeeping data across controllers
URL: https://github.com/apache/incubator-openwhisk/pull/2531#issuecomment-330598088
 
 
   @vvraskin to your points: 
   1 - in controller immediately, no. In controller in the near future, yes, e.g. in a mesos case we will need to use either ClusterSingleton or MemberEvent to designate a leader; we will also need to share some state within the cluster (may be able to leverage SharedDataService, or else directly use DistributedData)
   2 - my concern with downing is the case where nodes are down, and not coming back (or not coming back with the same id). In this case, docs indicate that new members cannot become "up", etc, till the unreachable node is downed. So regardless of whether traffic is routed to an unreachable node, if clustering usage is affected by having an unreachable, but NOT downed node, then there will be other types of problems (e.g. can replacement nodes be added? can the cluster size be grown?). Separately from needing the ability to cause nodes to be downed, the semantics of determining down vs *temp* unreachable will be different in a cluster manager like mesos/kube, similar to the semantics of determining seed nodes.   
 
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services