incubator-cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Nikolay Mihaylov <n...@nmmm.nu>
Subject Re: unable to delete
Date Fri, 07 Jun 2013 14:48:18 GMT
Hi

please note that when you drop column family, the data on the disk is not
deleted.

this is something you should do yourself.

>> Do the files get deleted on GC/server restart?
the question actually translates - do the column family existed after the
restart?

John pls correct me if I am explaining it wrong.

Nick.


On Mon, Jun 3, 2013 at 10:14 PM, Robert Coli <rcoli@eventbrite.com> wrote:

> On Mon, Jun 3, 2013 at 11:57 AM, John R. Frank <jrf@mit.edu> wrote:
> > Is it considered normal for cassandra to experience this error:
> >
> > ERROR [NonPeriodicTasks:1] 2013-06-03 18:17:05,374
> SSTableDeletingTask.java
> > (line 72) Unable to delete
> > /raid0/cassandra/data/<KEYSPACE>/<CF>/<KEYSPACE>-<CF>-ic-19-Data.db
(it
> will
> > be removed on server restart; we'll also retry after GC)
>
>
> cassandra//src/java/org/apache/cassandra/io/sstable/SSTableDeletingTask.java
> "
>        File datafile = new File(desc.filenameFor(Component.DATA));
>         if (!datafile.delete())
>         {
>             logger.error("Unable to delete " + datafile + " (it will
> be removed on server restart; we'll also retry after GC)");
>             failedTasks.add(this);
>             return;
>         }
> "
>
> There are contexts where it is appropriate for Cassandra to be unable
> to delete a file using io.File.delete.
>
> "
> // Deleting sstables is tricky because the mmapping might not have
> been finalized yet,
> // and delete will fail (on Windows) until it is (we only force the
> unmapping on SUN VMs).
> // Additionally, we need to make sure to delete the data file first,
> so on restart the others
> // will be recognized as GCable.
> "
>
> Do the files get deleted on GC/server restart?
>
> > This is on the DataStax EC2 AMI in a two-node cluster.  After deleting
> 1,000
> > rows from a CF with 20,000 rows, the DB becomes slow, and I'm trying to
> > figure out why.  Could this error message be pointing at a proximate
> cause?
>
> Almost certainly not. By the time that a sstable file is subject to
> deletion, it should no longer be "live". When it is no longer "live"
> it is not in the read path.
>
> You can verify this by using nodetool getsstables on a given key.
>
> What operation are you trying to do when the "DB becomes slow"?
>
> =Rob
>

Mime
View raw message