You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@cassandra.apache.org by Eunsu Kim <eu...@gmail.com> on 2019/09/05 08:51:29 UTC
about remaining data after adding a node
Hi, all
After adding a new node, all the data was streamed by the newly allocated token.
Since nodetool cleanup has not yet been performed on existing nodes, the total size has increased.
All data has a short ttl. In this case, will the data remaining on the existing node be deleted after the end of life? Or should I run nodetool cleanup to delete it?
Thanks in advance.
Re: about remaining data after adding a node
Posted by Alexander Dejanovski <al...@thelastpickle.com>.
Hi,
I advise not to run nodetool compact on a TWCS table.
If you do not want to run cleanup and are fine with the extra load on disk
for now, you can wait for data to expire naturally.
It will delete both data that is still owned by the nodes as well as data
that they don't own anymore once the sstables are fully expired.
Cheers,
-----------------
Alexander Dejanovski
France
@alexanderdeja
Consultant
Apache Cassandra Consulting
http://www.thelastpickle.com
On Thu, Sep 5, 2019 at 11:33 AM Eunsu Kim <eu...@gmail.com> wrote:
> Thank you for your response.
>
>
>
> I’m using TimeWindowCompactionStrategy.
>
>
>
> So if I don't run *nodetool compact*, will the remaining data not be
> deleted?
>
>
>
> *From: *Federico Razzoli <fe...@gmail.com>
> *Reply-To: *"user@cassandra.apache.org" <us...@cassandra.apache.org>
> *Date: *Thursday, 5 September 2019 at 6:19 PM
> *To: *"user@cassandra.apache.org" <us...@cassandra.apache.org>
> *Subject: *Re: about remaining data after adding a node
>
>
>
> Hi Eunsu,
>
>
>
> Are you using DateTieredCompactionStrategy? It optimises the deletion of
> expired data from disks.
>
> If minor compactions are not solving the problem, I suggest to run
> nodetool compact.
>
>
>
> Federico
>
>
>
>
>
> On Thu, 5 Sep 2019 at 09:51, Eunsu Kim <eu...@gmail.com> wrote:
>
>
>
> Hi, all
>
>
>
>
>
> After adding a new node, all the data was streamed by the newly allocated
> token.
>
>
>
>
>
> Since *nodetool cleanup* has not yet been performed on existing nodes,
> the total size has increased.
>
>
>
>
>
> All data has a short ttl. In this case, will the data remaining on the
> existing node be deleted after the end of life? Or should I run *nodetool
> cleanup* to delete it?
>
>
>
>
>
> Thanks in advance.
>
>
Re: about remaining data after adding a node
Posted by Eunsu Kim <eu...@gmail.com>.
Thank you for your response.
I’m using TimeWindowCompactionStrategy.
So if I don't run nodetool compact, will the remaining data not be deleted?
From: Federico Razzoli <fe...@gmail.com>
Reply-To: "user@cassandra.apache.org" <us...@cassandra.apache.org>
Date: Thursday, 5 September 2019 at 6:19 PM
To: "user@cassandra.apache.org" <us...@cassandra.apache.org>
Subject: Re: about remaining data after adding a node
Hi Eunsu,
Are you using DateTieredCompactionStrategy? It optimises the deletion of expired data from disks.
If minor compactions are not solving the problem, I suggest to run nodetool compact.
Federico
On Thu, 5 Sep 2019 at 09:51, Eunsu Kim <eu...@gmail.com>> wrote:
Hi, all
After adding a new node, all the data was streamed by the newly allocated token.
Since nodetool cleanup has not yet been performed on existing nodes, the total size has increased.
All data has a short ttl. In this case, will the data remaining on the existing node be deleted after the end of life? Or should I run nodetool cleanup to delete it?
Thanks in advance.
Re: about remaining data after adding a node
Posted by guo Maxwell <cc...@gmail.com>.
his data got ttl，so just wait if he do not want do cleanup
Oleksandr Shulgin <ol...@zalando.de>于2019年9月5日 周四下午5:48写道：
> On Thu, Sep 5, 2019 at 11:19 AM Federico Razzoli <
> federico.razzoli.dba@gmail.com> wrote:
>
>>
>> Are you using DateTieredCompactionStrategy? It optimises the deletion of
>> expired data from disks.
>> If minor compactions are not solving the problem, I suggest to run
>> nodetool compact.
>>
>
> Sorry, but both of the suggestions above are of dubious quality IMO.
>
> Don't use DTCS, use TWCS instead, if your use case is a good fit
> (immutable, in-order, TTLd inserts).
> Don't trigger major compaction for w/o a very good reason.
>
> In general, compaction can help to get rid of expired data, but it doesn't
> remove copies of data for which the node is not responsible anymore due to
> topology change. Use the cleanup command for that.
>
> --
> Alex
>
> --
you are the apple of my eye !
Re: about remaining data after adding a node
Posted by Oleksandr Shulgin <ol...@zalando.de>.
On Thu, Sep 5, 2019 at 11:19 AM Federico Razzoli <
federico.razzoli.dba@gmail.com> wrote:
>
> Are you using DateTieredCompactionStrategy? It optimises the deletion of
> expired data from disks.
> If minor compactions are not solving the problem, I suggest to run
> nodetool compact.
>
Sorry, but both of the suggestions above are of dubious quality IMO.
Don't use DTCS, use TWCS instead, if your use case is a good fit
(immutable, in-order, TTLd inserts).
Don't trigger major compaction for w/o a very good reason.
In general, compaction can help to get rid of expired data, but it doesn't
remove copies of data for which the node is not responsible anymore due to
topology change. Use the cleanup command for that.
--
Alex
Re: about remaining data after adding a node
Posted by Federico Razzoli <fe...@gmail.com>.
Hi Eunsu,
Are you using DateTieredCompactionStrategy? It optimises the deletion of
expired data from disks.
If minor compactions are not solving the problem, I suggest to run nodetool
compact.
Federico
On Thu, 5 Sep 2019 at 09:51, Eunsu Kim <eu...@gmail.com> wrote:
>
>
> Hi, all
>
>
>
>
>
> After adding a new node, all the data was streamed by the newly allocated
> token.
>
>
>
>
>
> Since *nodetool cleanup* has not yet been performed on existing nodes,
> the total size has increased.
>
>
>
>
>
> All data has a short ttl. In this case, will the data remaining on the
> existing node be deleted after the end of life? Or should I run *nodetool
> cleanup* to delete it?
>
>
>
>
>
> Thanks in advance.
>