[ 
https://issues.apache.org/jira/browse/BOOKKEEPER-1105?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Sijie Guo resolved BOOKKEEPER-1105.
-----------------------------------
       Resolution: Fixed
    Fix Version/s: 4.5.1
                   4.6.0

Issue resolved by merging pull request 425
            [https://github.com/apache/bookkeeper/pull/425]

            {noformat}
            commit f5d2ba66d6253fb54811d3ceb40d4ef4591dfb20
Author:     Andrey Yegorov <[email protected]>
AuthorDate: Tue Aug 15 18:41:49 2017 -0700
Commit:     Sijie Guo <[email protected]>
CommitDate: Tue Aug 15 18:41:49 2017 -0700

    BOOKKEEPER-1105: RackAwarePolicy: Failure to map node into rack may result 
in failure to add other nodes.
    
    - RackAwarePolicy's no longer uses /default-region if rack mapping fails 
unless required (by RegionAwarePolicy)
    - it no longer fails to add rest of nodes after one node's failed addition,
    - added unit tests
    - added counters for successful/failed bookie adds/removal
    (PR description content here)...
    
    UpdateLedgerOpTest failed but it seems to be known/unrelated issue.
    
    Author: Andrey Yegorov <[email protected]>
    
    Reviewers: Sijie Guo <[email protected]>
    
    This closes #425 from dlg99/fix/rackaware

            {noformat}
            

> RackAwarePolicy: Failure to map node into rack may result in failure to add 
> other nodes.
> ----------------------------------------------------------------------------------------
>
>                 Key: BOOKKEEPER-1105
>                 URL: https://issues.apache.org/jira/browse/BOOKKEEPER-1105
>             Project: Bookkeeper
>          Issue Type: Bug
>          Components: bookkeeper-client, bookkeeper-server
>    Affects Versions: 4.5.0
>            Reporter: Andrey Yegorov
>            Assignee: Andrey Yegorov
>            Priority: Critical
>             Fix For: 4.6.0, 4.5.1
>
>
> Ran into this issue:
> due to some misconfiguration of new node and rack assignments we've ended up 
> with nodes being assigned to default rack. We are not using region-aware, 
> only rack-aware policy.
> This generates the following sequence of events:
> - default rack is /default-region/default-rack even though region-aware is 
> not used.
> - other nodes mapped to racks like /rack1, /rack2 etc.
> - mixing /region/rack and /rack styles is not allowed, but exception on 
> addition of such nodes is swallowed. All following nodes to add (if there 
> were any) just skipped as result.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to