accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chris McCubbin (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (ACCUMULO-388) Agitator times should be ranges
Date Wed, 03 Oct 2012 21:39:07 GMT

     [ https://issues.apache.org/jira/browse/ACCUMULO-388?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Chris McCubbin updated ACCUMULO-388:
------------------------------------

    Assignee: Chris McCubbin
      Status: Patch Available  (was: Open)

Attached is a patch that alters the agitator script to allow for ranges on the input. current
input is like:

agitator.pl 1 2 3 4

Where 1 is the sleep time in minutes between kills and tup, and 2 is the sleep time between
tup and kills. 3 and 4 are minkill and maxkill resp.

The patch is fully backwards compatible in input and output. However, it now has the capability
to be called like this:

agitator.pl 1:5 2:6 3 4

where now the time between kill and tup will be a uniformly randomly distributed number between
1 and 5 inclusive, and similar with the tup to kill time. The two styles can be mixed:

agitator.pl 1:5 2 3 4

is also valid. The separating character must be a colon. The script will throw an error if
called with a minimum time greater than the max time.
                
> Agitator times should be ranges
> -------------------------------
>
>                 Key: ACCUMULO-388
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-388
>             Project: Accumulo
>          Issue Type: Improvement
>            Reporter: Keith Turner
>            Assignee: Chris McCubbin
>             Fix For: 1.5.0
>
>         Attachments: accumulo388patch.diff
>
>
> Currently the agitation script has a configurable time for time between kills and time
between restarts.  Currently these are single value times.  The times should be ranges with
the expectation that the script will pick a random number in that range.  This has the advantage
that sometimes a kill event will occurr while logs are still recovering from a previous kill
event and sometimes it will be after.  The same applies to the restart times, sometimes a
killed server will be restarted before the system has recovered from the previous kill and
sometimes after.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message