[ 
https://issues.apache.org/jira/browse/ZOOKEEPER-368?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12779178#action_12779178
 ] 

Henry Robinson commented on ZOOKEEPER-368:
------------------------------------------

Thanks Ben! I agree that 581 is a genuine issue, I'll take it up over on its 
JIRA. 

Mahadev - 

An Observer won't be able to connect to a pre-Observer ensemble because it 
doesn't send FOLLOWERINFO (rather, it sends OBSERVERINFO). The effect is that 
it retries, and is rejected. I have just verified this. 

If a server is brought down and reconnects as an Observer, it will be able to 
connect to the ensemble without problem. The Leader does not validate the type 
of the Learner that connects so it happily accepts the OBSERVERINFO handshake 
and carries on. It is possible that, if the process was restarted very quickly 
and the server was originally the Leader, that there might be some confusion 
when the Observer refuses to issue proposals. My belief is that the old Leader 
would be identified as failed. 

This should probably be considered user error? Users must not try and start the 
cluster with different configurations at each node. I can think of similar 
'bugs' in the current code where different servers have different 
configurations and therefore acknowledge different quorum groups, meaning that 
there wouldn't be consensus on who is the Leader, for example. 

> Observers: core functionality 
> ------------------------------
>
>                 Key: ZOOKEEPER-368
>                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-368
>             Project: Zookeeper
>          Issue Type: New Feature
>          Components: quorum
>            Reporter: Flavio Paiva Junqueira
>            Assignee: Henry Robinson
>             Fix For: 3.3.0
>
>         Attachments: obs-refactor.patch, observer-refactor.patch, observers 
> sync benchmark.png, observers.patch, ZOOKEEPER-368.patch, 
> ZOOKEEPER-368.patch, ZOOKEEPER-368.patch, ZOOKEEPER-368.patch, 
> ZOOKEEPER-368.patch, ZOOKEEPER-368.patch, ZOOKEEPER-368.patch, 
> ZOOKEEPER-368.patch, ZOOKEEPER-368.patch, ZOOKEEPER-368.patch
>
>
> Edit (Henry Robinson/henryr) 12/11/09:
> This JIRA specifically concerns the implementation of non-voting peers called 
> Observers, their documentation and their tests. 
> Explicit goals are 1. not breaking any current ZK functionality, 2. enabling 
> at least one deployment scenario involving Observers, 3. documentation 
> describing how to use the feature and 4. tests validating the correct 
> behaviour of 2. 
> Non goals of this JIRA are 1. performance optimizations specific to 
> Observers, 2. compatibility with every feature of ZooKeeper (in particular 
> all leader election protocols), which are both to be addressed in future 
> JIRAs. 
> See http://wiki.apache.org/hadoop/ZooKeeper/Observers for more detail of use 
> cases, proposed design and usage.
> See http://wiki.apache.org/hadoop/ZooKeeper/Observers/ReviewGuide for a brief 
> commentary on the current patch. 
> -------------
> Currently, all servers of an ensemble participate actively in reaching 
> agreement on the order of ZooKeeper transactions. That is, all followers 
> receive proposals, acknowledge them, and receive commit messages from the 
> leader. A leader issues commit messages once it receives acknowledgments from 
> a quorum of followers. For cross-colo operation, it would be useful to have a 
> third role: observer. Using Paxos terminology, observers are similar to 
> learners. An observer does not participate actively in the agreement step of 
> the atomic broadcast protocol. Instead, it only commits proposals that have 
> been accepted by some quorum of followers.
> One simple solution to implement observers is to have the leader forwarding 
> commit messages not only to followers but also to observers, and have 
> observers applying transactions according to the order followers agreed upon. 
> In the current implementation of the protocol, however, commit messages do 
> not carry their corresponding transaction payload because all servers 
> different from the leader are followers and followers receive such a payload 
> first through a proposal message. Just forwarding commit messages as they 
> currently are to an observer consequently is not sufficient. We have a couple 
> of options:
> 1- Include the transaction payload along in commit messages to observers;
> 2- Send proposals to observers as well.
> Number 2 is simpler to implement because it doesn't require changing the 
> protocol implementation, but it increases traffic slightly. The performance 
> impact due to such an increase might be insignificant, though.
> For scalability purposes, we may consider having followers also forwarding 
> commit messages to observers. With this option, observers can connect to 
> followers, and receive messages from followers. This choice is important to 
> avoid increasing the load on the leader with the number of observers. 

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.

Reply via email to