[ 
https://issues.apache.org/jira/browse/GEODE-8394?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17173333#comment-17173333
 ] 

ASF subversion and git services commented on GEODE-8394:
--------------------------------------------------------

Commit 83d1e28a953b7d73e7f499f9013540bedd0bd472 in geode's branch 
refs/heads/develop from agingade
[ https://gitbox.apache.org/repos/asf?p=geode.git;h=83d1e28 ]

GEODE-8394: Rewind the message Part on command failure (#5424)

GEODE-8394: Rewind the message Part on failure

Co-authored-by: anilkumar gingade <[email protected]>

> Client sends partial data during retry
> --------------------------------------
>
>                 Key: GEODE-8394
>                 URL: https://issues.apache.org/jira/browse/GEODE-8394
>             Project: Geode
>          Issue Type: Bug
>          Components: client/server
>    Affects Versions: 1.14.0
>            Reporter: Anilkumar Gingade
>            Assignee: Anilkumar Gingade
>            Priority: Major
>              Labels: GeodeOperationAPI, pull-request-available
>
> When executing the command, java client sends the command in the form of 
> message (Parts). If there is a failure during the send, it retries number of 
> times based on the retry-attempt setting.
> If there is a failure in the first attempt due to Exception (IOException, 
> EOFException - read timeout); during the second attempt the client is sending 
> partial data, instead of the complete data.
> This could be reproduced by setting a small read-timeout and large object 
> (Put).
> Because of this, the put from client succeeds by creating partial data on the 
> server cache. Since the data is stored in serialized form, there is no 
> exception thrown during put() operation, giving an impression that put 
> operation is successful. 
> The workaround for now is to have large read-timeout setting; while trying to 
> send a large object.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to