[jboss-jira] [JBoss JIRA] (JGRP-2472) TCPConnection allocates 1.5G after reading incorrect header in the connection

[Pavel Molchanov (Jira)]

     [ https://issues.redhat.com/browse/JGRP-2472?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Pavel Molchanov updated JGRP-2472:
----------------------------------
    Attachment: 2020-04-21_12h58_50.png
                2020-04-21_15h38_37.png


> TCPConnection allocates 1.5G after reading incorrect header in the connection
> -----------------------------------------------------------------------------
>
>                 Key: JGRP-2472
>                 URL: https://issues.redhat.com/browse/JGRP-2472
>             Project: JGroups
>          Issue Type: Bug
>    Affects Versions: 3.6.13
>            Reporter: Pavel Molchanov
>            Assignee: Bela Ban
>            Priority: Major
>         Attachments: 2020-04-21_12h58_50.png, 2020-04-21_15h38_37.png
>
>
> We are using FILE_PING JGroups functionality with Artemis servers.
> Periodically (very rate) we see that TCPConnection allocates 1650814049 bytes in the receiving buffer.
> This is similar to the https://issues.redhat.com/browse/JGRP-1117 issue.
> This issue is happening periodically and it's clearly visible in the memory dumps from Artemis servers.
> Screen-shots of the dumps are attached.
> Can TCPConnection be protected from the incorrect input?
> https://github.com/belaban/JGroups/blob/JGroups-3.6.13.Final/src/org/jgroups/blocks/cs/TcpConnection.java
> Something like this:
> {code}
> int len=in.readInt();  
> if(len==1650814049) throw InvalidStateException(); 
> {code}



--
This message was sent by Atlassian Jira
(v7.13.8#713008)