jclouds-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF subversion and git services (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (JCLOUDS-460) Add random delay to avoid thundering herd problem
Date Tue, 04 Mar 2014 06:31:24 GMT

    [ https://issues.apache.org/jira/browse/JCLOUDS-460?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13919053#comment-13919053
] 

ASF subversion and git services commented on JCLOUDS-460:
---------------------------------------------------------

Commit 00847f9ffc5fade4e63530b79c7d38962c1de08f in jclouds's branch refs/heads/master from
[~gaul]
[ https://git-wip-us.apache.org/repos/asf?p=jclouds.git;h=00847f9 ]

JCLOUDS-460: Add jitter to avoid thundering herd

When issuing many simultaneous requests to Synaptic Atmos I observed:

HTTP/1.1 failed with code 500, error: AtmosError
[code=1040, message=The server is busy. Please try again.]

Previously all clients slept for fixed intervals and thus retried
around the same time.  This commit adds a random delay which should
better distribute load on the provider.


> Add random delay to avoid thundering herd problem
> -------------------------------------------------
>
>                 Key: JCLOUDS-460
>                 URL: https://issues.apache.org/jira/browse/JCLOUDS-460
>             Project: jclouds
>          Issue Type: Improvement
>          Components: jclouds-blobstore
>    Affects Versions: 1.7.1
>            Reporter: Andrew Gaul
>            Assignee: Andrew Gaul
>
> When issuing many simultaneous requests to Synaptic Atmos I observed:
> HTTP/1.1 failed with code 500, error: AtmosError [code=1040, message=The server is busy.
Please try again.]
> This is a general problem for private blobstores since a single client can often overwhelm
the server with many requests.  While jclouds retries in many of these situations, it waits
a fixed amount of time between requests, often overwhelming the server again.  Instead it
should insert some jitter, e.g., 0-10% extra delay, to more evenly distribute the retries.



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

Mime
View raw message