You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@geode.apache.org by "ASF subversion and git services (JIRA)" <ji...@apache.org> on 2017/05/18 17:00:09 UTC

[jira] [Commented] (GEODE-2804) Allow a locator host to be taken off line and replaced with a different machine having the same host name

    [ https://issues.apache.org/jira/browse/GEODE-2804?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16016075#comment-16016075 ] 

ASF subversion and git services commented on GEODE-2804:
--------------------------------------------------------

Commit e216fde1e4b1613bde22112cfb1544be022c3aac in geode's branch refs/heads/develop from [~hitesh.khamesra]
[ https://git-wip-us.apache.org/repos/asf?p=geode.git;h=e216fde ]

GEODE-2804 Update InetSocketAddress, when there is IOException.

Geode keeps InetSocketAddress for locators. So, if locators ip
changes in cloud/VM enviroment then Geode process unable to
connect to locator. Thus we have fixed this problem in two way.

a. If Geode client sees IOException while connecting to locator then
we change cached InetAddress to use locator hostname. In this way,
client does DNS query again for locator host.
b. For other Geode process, now we connect to locator using hostname.

Added couple of junit tests for it.


> Allow a locator host to be taken off line and replaced with a different machine having the same host name
> ---------------------------------------------------------------------------------------------------------
>
>                 Key: GEODE-2804
>                 URL: https://issues.apache.org/jira/browse/GEODE-2804
>             Project: Geode
>          Issue Type: Improvement
>          Components: client/server, membership, native client
>    Affects Versions: 1.1.1
>            Reporter: Bruce Schuchardt
>
> It is not unlikely that in a long-running system a machine hosting a locator will break down and need to be replace, or that a virtual machine hosting a locator will be stopped and restarted.  In either case the machine may have a different IP address.
> Clients and servers currently cache the IP address of a locator and should be changed to re-resolve the host name if the address stops being usable.
> Servers currently cannot be started if they are given a host name that can not be resolved.  They should be changed to allow this and start up if they are able to contact one or more of the other locators in their "locators" setting.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)