tephra-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (TEPHRA-227) DebugTool - show regions that have not yet recorded prune upper bound
Date Thu, 02 Mar 2017 22:17:45 GMT

    [ https://issues.apache.org/jira/browse/TEPHRA-227?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15893142#comment-15893142

ASF GitHub Bot commented on TEPHRA-227:

Github user poornachandra commented on a diff in the pull request:

    --- Diff: tephra-hbase-compat-0.96/src/main/java/org/apache/tephra/hbase/txprune/InvalidListPruningDebug.java
    @@ -148,6 +232,9 @@ private void printUsage(PrintWriter pw) {
                      "provided as the limit, prune upper bounds of all regions are returned.");
         pw.println("prune-info region-name-as-string");
         pw.println("Desc: Prints out the Pruning information for the region 'region-name-as-string'");
    +    pw.println("to-compact-regions limit");
    +    pw.println("Desc: Prints out 'limit' number of regions that are active, but are not
empty, " +
    +                 "but have not registered a prune upper bound.");
    --- End diff --
    `and have not registered a prune upper bound.`

> DebugTool - show regions that have not yet recorded prune upper bound 
> ----------------------------------------------------------------------
>                 Key: TEPHRA-227
>                 URL: https://issues.apache.org/jira/browse/TEPHRA-227
>             Project: Tephra
>          Issue Type: Improvement
>    Affects Versions: 0.11.0-incubating
>            Reporter: Gokul Gunasekaran
>            Assignee: Poorna Chandra
>             Fix For: 0.11.0-incubating
> i) getIdleRegions (numRegions) --> since this method returns all the regions that
have ever existed, this is not very useful since some regions might not be existing anymore.
So, we need to intersect the idleregions with the latest TimeRegion regions.
> ii) add a new method - getRegionsToBeFlushed ( ) --> that returns the regions that
are part of the latest TimeRegion but don't have any prune upper bound recorded and or not
empty either.
> So during debugging of prune progress, first we can hit the second method and get the
regions to be flushed/compacted (since any regions that have not compacted/pruned will prevent
pruning from progression). After that list is empty, we will hit the first endpoint to find
what regions are lagging, pruning/compacting them will help advance pruning.

This message was sent by Atlassian JIRA

View raw message