You are viewing a plain text version of this content. The canonical link for it is here.
Posted to yarn-issues@hadoop.apache.org by "Karthik Kambatla (JIRA)" <ji...@apache.org> on 2013/10/17 21:42:46 UTC
[jira] [Updated] (YARN-1222) Make improvements in ZKRMStateStore
for fencing
[ https://issues.apache.org/jira/browse/YARN-1222?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Karthik Kambatla updated YARN-1222:
-----------------------------------
Attachment: yarn-1222-2.patch
Here is an updated patch on recent trunk. I am yet to add unit tests for the patch, posting it for a high-level design review.
High-level approach:
# Every write operation is padded between creation and deletion of a fencing node.
# The root node is governed by special ACLs that gives the RM exclusive create-delete access, and shared read-write-admin.
# Client can set ACLs for the root node. If client doesn't explicitly set, the default fencing mode kicks in - we generate the root-node ACLs from the store-ACLs by stripping create-delete perms of all users but the RM. RM identifies itself through the RM_ADDRESS and clusterTimeStamp.
> Make improvements in ZKRMStateStore for fencing
> -----------------------------------------------
>
> Key: YARN-1222
> URL: https://issues.apache.org/jira/browse/YARN-1222
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Bikas Saha
> Assignee: Karthik Kambatla
> Attachments: yarn-1222-1.patch, yarn-1222-2.patch
>
>
> Using multi-operations for every ZK interaction.
> In every operation, automatically creating/deleting a lock znode that is the child of the root znode. This is to achieve fencing by modifying the create/delete permissions on the root znode.
--
This message was sent by Atlassian JIRA
(v6.1#6144)