You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@cassandra.apache.org by kevin <ke...@gmail.com> on 2010/01/24 08:31:34 UTC
disk space and load after removing data
after deleting a lot of column families, column the disk space used by
cassandra and the Load reported by nodeprobe info is not going down at all.
is there something i have to do recover the disk space? also what about the
Load reported by nodeprobe info?
thanks
Re: disk space and load after removing data
Posted by Jonathan Ellis <jb...@gmail.com>.
I (finally) wrote a blog post about this:
http://spyced.blogspot.com/2010/02/distributed-deletes-in-cassandra.html
I've also put it on the wiki in slightly modified form:
http://wiki.apache.org/cassandra/DistributedDeletes
On Sun, Jan 24, 2010 at 1:31 AM, kevin <ke...@gmail.com> wrote:
> after deleting a lot of column families, column the disk space used by
> cassandra and the Load reported by nodeprobe info is not going down at all.
>
> is there something i have to do recover the disk space? also what about the
> Load reported by nodeprobe info?
> thanks
>
Re: disk space and load after removing data
Posted by Ramzi Rabah <rr...@playdom.com>.
Need to restart I believe
On Mon, Jan 25, 2010 at 7:34 AM, kevin <ke...@gmail.com> wrote:
> thanks a lot, the default seems 10 days! do i have to stop and start
> cassandra or is there some command to make it reload the new settings?
> thanks
>
> On Sun, Jan 24, 2010 at 10:36 PM, Ramzi Rabah <rr...@playdom.com> wrote:
>>
>> GCGraceSeconds is the amount of seconds the system waits before it
>> gets rids of tombstone records. To get rid of them faster and release
>> the space taken by them change it in the storage config file to a
>> smaller value (like 3600)
>>
>> On Sun, Jan 24, 2010 at 9:40 PM, kevin <ke...@gmail.com> wrote:
>> > any help here guys?
>> > thanks a lot!
>> >
>> > On Sun, Jan 24, 2010 at 10:51 AM, kevin <ke...@gmail.com>
>> > wrote:
>> >>
>> >> i did
>> >> nodeprobe -host localhost flush Keyspace1
>> >> nodeprobe -host localhost compact
>> >> nodeprobe -host localhost cleanup
>> >>
>> >>
>> >> and there is no change in the disk usage or Load info
>> >>
>> >> On Sat, Jan 23, 2010 at 11:31 PM, kevin <ke...@gmail.com>
>> >> wrote:
>> >>>
>> >>> after deleting a lot of column families, column the disk space used by
>> >>> cassandra and the Load reported by nodeprobe info is not going down at
>> >>> all.
>> >>>
>> >>> is there something i have to do recover the disk space? also what
>> >>> about
>> >>> the Load reported by nodeprobe info?
>> >>> thanks
>> >>
>> >
>> >
>
>
Re: disk space and load after removing data
Posted by kevin <ke...@gmail.com>.
thanks a lot, the default seems 10 days! do i have to stop and start
cassandra or is there some command to make it reload the new settings?
thanks
On Sun, Jan 24, 2010 at 10:36 PM, Ramzi Rabah <rr...@playdom.com> wrote:
> GCGraceSeconds is the amount of seconds the system waits before it
> gets rids of tombstone records. To get rid of them faster and release
> the space taken by them change it in the storage config file to a
> smaller value (like 3600)
>
> On Sun, Jan 24, 2010 at 9:40 PM, kevin <ke...@gmail.com> wrote:
> > any help here guys?
> > thanks a lot!
> >
> > On Sun, Jan 24, 2010 at 10:51 AM, kevin <ke...@gmail.com>
> wrote:
> >>
> >> i did
> >> nodeprobe -host localhost flush Keyspace1
> >> nodeprobe -host localhost compact
> >> nodeprobe -host localhost cleanup
> >>
> >>
> >> and there is no change in the disk usage or Load info
> >>
> >> On Sat, Jan 23, 2010 at 11:31 PM, kevin <ke...@gmail.com>
> >> wrote:
> >>>
> >>> after deleting a lot of column families, column the disk space used by
> >>> cassandra and the Load reported by nodeprobe info is not going down at
> all.
> >>>
> >>> is there something i have to do recover the disk space? also what about
> >>> the Load reported by nodeprobe info?
> >>> thanks
> >>
> >
> >
>
Re: disk space and load after removing data
Posted by Ramzi Rabah <rr...@playdom.com>.
GCGraceSeconds is the amount of seconds the system waits before it
gets rids of tombstone records. To get rid of them faster and release
the space taken by them change it in the storage config file to a
smaller value (like 3600)
On Sun, Jan 24, 2010 at 9:40 PM, kevin <ke...@gmail.com> wrote:
> any help here guys?
> thanks a lot!
>
> On Sun, Jan 24, 2010 at 10:51 AM, kevin <ke...@gmail.com> wrote:
>>
>> i did
>> nodeprobe -host localhost flush Keyspace1
>> nodeprobe -host localhost compact
>> nodeprobe -host localhost cleanup
>>
>>
>> and there is no change in the disk usage or Load info
>>
>> On Sat, Jan 23, 2010 at 11:31 PM, kevin <ke...@gmail.com>
>> wrote:
>>>
>>> after deleting a lot of column families, column the disk space used by
>>> cassandra and the Load reported by nodeprobe info is not going down at all.
>>>
>>> is there something i have to do recover the disk space? also what about
>>> the Load reported by nodeprobe info?
>>> thanks
>>
>
>
Re: disk space and load after removing data
Posted by kevin <ke...@gmail.com>.
any help here guys?
thanks a lot!
On Sun, Jan 24, 2010 at 10:51 AM, kevin <ke...@gmail.com> wrote:
> i did
> nodeprobe -host localhost flush Keyspace1
> nodeprobe -host localhost compact
> nodeprobe -host localhost cleanup
>
>
> and there is no change in the disk usage or Load info
>
>
> On Sat, Jan 23, 2010 at 11:31 PM, kevin <ke...@gmail.com>wrote:
>
>> after deleting a lot of column families, column the disk space used by
>> cassandra and the Load reported by nodeprobe info is not going down at all.
>>
>> is there something i have to do recover the disk space? also what about
>> the Load reported by nodeprobe info?
>> thanks
>>
>
>
Re: disk space and load after removing data
Posted by kevin <ke...@gmail.com>.
i did
nodeprobe -host localhost flush Keyspace1
nodeprobe -host localhost compact
nodeprobe -host localhost cleanup
and there is no change in the disk usage or Load info
On Sat, Jan 23, 2010 at 11:31 PM, kevin <ke...@gmail.com> wrote:
> after deleting a lot of column families, column the disk space used by
> cassandra and the Load reported by nodeprobe info is not going down at all.
>
> is there something i have to do recover the disk space? also what about the
> Load reported by nodeprobe info?
> thanks
>