accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Keith Turner (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ACCUMULO-2608) Create a randomwalk replay meta-node
Date Wed, 16 Apr 2014 17:18:15 GMT

    [ https://issues.apache.org/jira/browse/ACCUMULO-2608?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13971674#comment-13971674
] 

Keith Turner commented on ACCUMULO-2608:
----------------------------------------

Some things to consider :

 * Test nodes make random decisions.  For example the shard Insert nodes generates random
documents.  So even if you visited notes in the Shard test in the same order, completely different
things could happen because different data was generated.  This could possibly be addressed
by using the same PRNG and seeding it the same. 
 * Concurrent walkers will generate multiple trails/paths.  Replaying those trail concurrently
will very likely not result in the same sequence of events globally.  
 * The XML file the ticket mentions generating and XML file.  This may not fit in memory.

 In my experience when I rerun random walk I usually see the same bugs.  If I only see a bug
once over weeks of running random walk on 20 nodes, it indicates that the bug may be improbable
and therefore of a lower priority.  There are certainly a few bugs that I only ever saw once
in random walk and could not figure them and never saw them again.  I saw these on a cluster
w/ many walkers.

> Create a randomwalk replay meta-node
> ------------------------------------
>
>                 Key: ACCUMULO-2608
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-2608
>             Project: Accumulo
>          Issue Type: New Feature
>          Components: test
>            Reporter: Mike Drob
>            Assignee: Bill Havanki
>            Priority: Critical
>             Fix For: 1.4.6, 1.5.2, 1.6.1
>
>
> It would be useful to have the ability to replay randomwalk sequences.
> If a randomwalk fails, it is often difficult to reproduce due to the random nature of
things. If current test runs were modified to output a specially crafted module representing
their sequence, then future runs could accept it as the module to run and improve reproduction
capabilities.



--
This message was sent by Atlassian JIRA
(v6.2#6252)

Mime
View raw message