giraph-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Pavan Kumar (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (GIRAPH-821) proper handling of NegativeArraySizeException for all ByteArray backed messagestores
Date Tue, 07 Jan 2014 20:07:51 GMT

    [ https://issues.apache.org/jira/browse/GIRAPH-821?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13864628#comment-13864628
] 

Pavan Kumar commented on GIRAPH-821:
------------------------------------

The patch for testing was not included in the original patch since it requires at least 2856MB
of memory, and the test will break if such amount of memory is not available on the client
machine

>  proper handling of NegativeArraySizeException for all ByteArray backed messagestores
> -------------------------------------------------------------------------------------
>
>                 Key: GIRAPH-821
>                 URL: https://issues.apache.org/jira/browse/GIRAPH-821
>             Project: Giraph
>          Issue Type: Improvement
>            Reporter: Pavan Kumar
>            Assignee: Pavan Kumar
>            Priority: Minor
>              Labels: giraph
>         Attachments: GIRAPH-821.patch, Testing.patch
>
>
> For byte array backed message stores (i.e.     LongByteArrayMessageStore, ByteArrayMessagesPerVertexMessageStore,
and IntByteArrayMessageStore)     currently, when the byte[] of underlying ExtendedDataOuput
runs out of space (i.e. UnsafeByteArrayOutputStream)     we see java.lang.NegativeArraySizeException
in the SendWorkerMessagesRequest.     We want to catch the exception dump the known vertex
information and explain the issue     (the amount of bytes sent to this vertex exceeded the
max capacity of this ExtendedDataOutput).      In this diff all ByteArray backed messagestores
have been modified to behave as explained.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)

Mime
View raw message