[
https://issues.apache.org/jira/browse/HBASE-14141?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15652672#comment-15652672
]
Vladimir Rodionov commented on HBASE-14141:
-------------------------------------------
{quote}
We also have to think about the failure case where a WAL will be left un-closed
in case of RS dead. We cannot rely on a mechanism to write data in WAL close
because it will never be reliable. Even if we do a solution where we keep track
of Tables/Regions in the WAL and retroactively write this info to the backup
metadata, we have to design the system so that WALs from RS failures are
handled.
{quote}
We already depend on WALs for incremental backup. If WAL is unreliable so is
HBase. Backup can't be more reliable than WAL/HBase.
{quote}
Let's say I have a single huge table in the cluster, and a single backup set.
This means that we cannot use multi-wal at all, making the design decision a
non-starter.
{quote}
No. In this case, default mode is the way to go (what we have now).
> HBase Backup/Restore Phase 3: Filter WALs on backup to include only edits
> from backup tables
> --------------------------------------------------------------------------------------------
>
> Key: HBASE-14141
> URL: https://issues.apache.org/jira/browse/HBASE-14141
> Project: HBase
> Issue Type: New Feature
> Affects Versions: 2.0.0
> Reporter: Vladimir Rodionov
> Assignee: Vladimir Rodionov
> Labels: backup
> Fix For: 2.0.0
>
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
