[
https://issues.apache.org/jira/browse/YARN-3736?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14632164#comment-14632164
]
Subru Krishnan commented on YARN-3736:
--------------------------------------
Thanks [~adhoot] for the patch.
I have a few initial comments:
* This patch includes the fix for YARN-3900. I see that [~jianhe] has
reviewed & ok-ed it so we should rebase this patch after committing it.
* _ReservationId_ is redudant in *ReservationAllocationStateProto* as that's
the key of the entry.
* It may be better to store _ReservationId_ object rather than the name as
we are parsing the name back in most places.
* In *LeveldbRMStateStore* we use batch writes for delete and not for
store/update.
* In *ZKRMStateStore*, do we need to use _SafeTransaction_ for deleting
reservations also.
> Persist the Plan information, ie. accepted reservations to the RMStateStore
> for failover
> ----------------------------------------------------------------------------------------
>
> Key: YARN-3736
> URL: https://issues.apache.org/jira/browse/YARN-3736
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: capacityscheduler, fairscheduler, resourcemanager
> Reporter: Subru Krishnan
> Assignee: Anubhav Dhoot
> Attachments: YARN-3736.001.patch, YARN-3736.001.patch
>
>
> We need to persist the current state of the plan, i.e. the accepted
> ReservationAllocations & corresponding RLESpareseResourceAllocations to the
> RMStateStore so that we can recover them on RM failover. This involves making
> all the reservation system data structures protobuf friendly.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
