hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Andrew Purtell (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HBASE-6696) Add CP hooks pre and post split transaction point-of-no-return
Date Sun, 11 May 2014 11:12:17 GMT

     [ https://issues.apache.org/jira/browse/HBASE-6696?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel

Andrew Purtell updated HBASE-6696:

    Assignee:     (was: Andrew Purtell)

> Add CP hooks pre and post split transaction point-of-no-return
> --------------------------------------------------------------
>                 Key: HBASE-6696
>                 URL: https://issues.apache.org/jira/browse/HBASE-6696
>             Project: HBase
>          Issue Type: Improvement
>          Components: Coprocessors, regionserver
>    Affects Versions: 0.94.2, 0.95.2
>            Reporter: Andrew Purtell
> In the discussion "Improving Coprocessor postSplit/postOpen synchronization" on user@hbase,
a user implementing a secondary indexing scheme writes in:
> {quote}
> The goal with splits is to create two new daughter regions with the
> corresponding splits of the secondary indexes and lock these regions such
> that Puts/Deletes that occur while postSplit is in progress will be queued
> up so we don't run into consistency issues. [...] As of right now, the HBase coprocessors
do not easily support a way to achieve this level of consistency in that there is no way to
distinguish a region being opened from a split or a regular open.
> {quote}
> Setting aside the particulars of the use case, the issue is the {{preSplit}} hook does
not provide the identities of the daughter regions (they don't exist yet) and the {{postSplit}}
hook, while providing the identities of the daughter regions, runs after the master has processed
the split and the daughters are already opened or opening. A CP implementer has no interception
point available where the daughters exist but are not yet open.
> This JIRA proposes to add two new CP hooks to just before and after the point-of-no-return
(PONR) in the split transaction: {{preSplitPONR}} and {{postSplitPONR}}. Perhaps the naming
can be improved. This will address the above use case but additionally support overloading
of the split transaction itself. We also need to address observer notification if the split
transaction fails. This is like HBASE-5827 but scoped to this specific consideration only.

This message was sent by Atlassian JIRA

View raw message