zookeeper-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Michael Han (JIRA)" <j...@apache.org>
Subject [jira] [Assigned] (ZOOKEEPER-3485) Measure reconfiguration time
Date Fri, 02 Aug 2019 04:12:00 GMT

     [ https://issues.apache.org/jira/browse/ZOOKEEPER-3485?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Michael Han reassigned ZOOKEEPER-3485:
--------------------------------------

    Assignee: Karolos Antoniadis

> Measure reconfiguration time
> ----------------------------
>
>                 Key: ZOOKEEPER-3485
>                 URL: https://issues.apache.org/jira/browse/ZOOKEEPER-3485
>             Project: ZooKeeper
>          Issue Type: Improvement
>    Affects Versions: 3.5.5
>            Reporter: Karolos Antoniadis
>            Assignee: Karolos Antoniadis
>            Priority: Minor
>
> This issue is created after some initial discussion in the _dev_ mailing list (subject
"Leader election logging during reconfiguration").
>  
> There does not seem to be a good way to measure reconfiguration time in ZooKeeper. Additionally, reconfiguration
time is mixed together with leader election time*.* For instance, during reconfiguration,
ZooKeeper logs a  {{LEADER ELECTION TOOK}} message even though no leader election might
takes place.
>   
>  This can be reproduced by following these steps:
>  1) start a ZooKeeper cluster (e.g., 3 participants)
>  2) start a client that connects to some follower
>  3) perform a _reconfig_ operation that removes the leader from the cluster
>   
>  After the reconfiguration takes place, we can see that the log files of the remaining
participants contain a "_LEADER ELECTION TOOK_" message. For example, a line that contains
>  _2019-07-29 23:07:38,518 [myid:2] - INFO  [QuorumPeer[myid=2](plain=0.0.0.0:2792)(secure=disabled):Follower@75]
- FOLLOWING - LEADER ELECTION TOOK - 57 MS_
>   
>  However, no leader election took place, in the sense that no server went _LOOKING_ and
then started voting and sending notifications to other participants as would be in a normal
leader election. It seems, that before the _reconfig_ is committed, the participant that
is going to be the next leader is already decided (see here: [https://github.com/apache/zookeeper/blob/master/zookeeper-server/src/main/java/org/apache/zookeeper/server/quorum/Leader.java#L865]).
>   
> *Goal* of this issue/improvement is to measure in a better and more accurate way the
time it takes for a reconfiguration to complete, as well as, to clearly distinguish the measurement
of reconfiguration versus leader election.
>   
>   
>   



--
This message was sent by Atlassian JIRA
(v7.6.14#76016)

Mime
View raw message