accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Josh Elser (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ACCUMULO-3601) master does not process migrations
Date Sat, 21 Feb 2015 00:20:12 GMT

    [ https://issues.apache.org/jira/browse/ACCUMULO-3601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14329827#comment-14329827
] 

Josh Elser commented on ACCUMULO-3601:
--------------------------------------

[~ctubbsii] made the same suggestion to me in IRC (I'm now hearing your voice in what he said).
If that's the case, the DeleteTable FATE op's CleanUp class isReady() isn't doing things right.
The CleanUp gets stuck in isReady(), repeatedly hitting the following:

{code}
    for (Entry<Key,Value> entry : scanner) {
      log.debug(WholeRowIterator.decodeRow(entry.getKey(), entry.getValue()));
      TabletLocationState locationState = MetaDataTableScanner.createTabletLocationState(entry.getKey(),
entry.getValue());
      if (!locationState.extent.isPreviousExtent(prevExtent)) {
        log.info("Still waiting for table to be deleted: " + tableId + " saw inconsistency
" + prevExtent + " " + locationState.extent);
        done = false;
        break;
      }
{code}

Currently wrapping my head around this to figure out what it's supposed to be doing instead
(and how the recent changes would have broken this).

> master does not process migrations
> ----------------------------------
>
>                 Key: ACCUMULO-3601
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-3601
>             Project: Accumulo
>          Issue Type: Bug
>          Components: master
>            Reporter: Eric Newton
>            Assignee: Eric Newton
>            Priority: Blocker
>             Fix For: 1.5.3, 1.7.0, 1.6.3
>
>          Time Spent: 1h 10m
>  Remaining Estimate: 0h
>
> While attempting to write a test for ACCUMULO-3597, I found the master was not balancing.
 The iterator that filters the metadata table does not take migrations into account, so the
state of the tablets was never seen, and processed.  This bug was introduced with the fix
to ACCUMULO-3580.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message