[jboss-jira] [JBoss JIRA] Commented: (JGRP-960) FILE_DISCOVERY: new discovery protocol

[Bela Ban (JIRA)]

    [ https://jira.jboss.org/jira/browse/JGRP-960?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12464112#action_12464112 ] 

Bela Ban commented on JGRP-960:
-------------------------------

Comment from Bob:
Is there any race conditions we need to look out for when working with S3?  Or do you need to write S3-based lock-manager?  With XA and a pony?   :)

Or a separate thought, treat S3 more directory-like, and instead of updating a file, each node puts (and removes, optionally) its own file within the dir?

s3://path/to/my/cluster/node1.conf
s3://path/to/my/cluster/node2.conf
s3://path/to/my/cluster/node3.conf
s3://path/to/my/cluster/node4.conf
s3://path/to/my/cluster/node5.conf

Just thinking if 100 nodes are booting up at the same time, each trying to write to a shared cluster.conf without trampling each other? 

> FILE_DISCOVERY: new discovery protocol
> --------------------------------------
>
>                 Key: JGRP-960
>                 URL: https://jira.jboss.org/jira/browse/JGRP-960
>             Project: JGroups
>          Issue Type: Feature Request
>            Reporter: Bela Ban
>            Assignee: Bela Ban
>             Fix For: 2.8
>
>
> Uses a shared file to discover existing nodes. A new node writes its address to the file and removes it on leaving the cluster. Old entries are aged out after some time

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: https://jira.jboss.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira