cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Marcus Eriksson (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (CASSANDRA-12208) Estimated droppable tombstones given by sstablemetadata counts tombstones that aren't actually "droppable"
Date Tue, 09 Aug 2016 08:39:22 GMT

     [ https://issues.apache.org/jira/browse/CASSANDRA-12208?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Marcus Eriksson updated CASSANDRA-12208:
----------------------------------------
    Fix Version/s: 3.9
                   3.0.9
           Status: Patch Available  (was: Open)

patch [here|https://github.com/krummas/cassandra/commits/marcuse/12208] that adds a command
line parameter to tools/bin/sstablemetadata to state the gc_grace_seconds to use when calculating
the droppable tombstone estimate

http://cassci.datastax.com/view/Dev/view/krummas/job/krummas-marcuse-12208-dtest/
http://cassci.datastax.com/view/Dev/view/krummas/job/krummas-marcuse-12208-testall/

> Estimated droppable tombstones given by sstablemetadata counts tombstones that aren't
actually "droppable"
> ----------------------------------------------------------------------------------------------------------
>
>                 Key: CASSANDRA-12208
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-12208
>             Project: Cassandra
>          Issue Type: Bug
>            Reporter: Thanh
>            Assignee: Marcus Eriksson
>            Priority: Minor
>             Fix For: 3.0.9, 3.9
>
>
> => "Estimated droppable tombstones" given by *sstablemetadata* counts tombstones that
aren't actually "droppable"
> To be clear, the "Estimated droppable tombstones" calculation counts tombstones that
have not yet passed gc_grace_seconds as droppable tombstones, which is unexpected, since such
tombstones aren't droppable.
> To observe the problem:
> Create a table using the default gc_grace_seconds (default gc_grace_seconds is 86400
is 1 day).
> Populate the table with a couple of records.
> Do a delete.
> Do a "nodetool flush" to flush the memtable to disk.
> Do an "sstablemetadata <sstable>" to get the metadata of the sstable you just created
by doing the flush, and observe that the Estimated droppable tombstones is greater than 0.0
(actual value depends on the total number inserts/updates/deletes that you did before triggered
the flush)



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

Mime
View raw message