Fwd: Questions about database file sizes.

[Sivan Greenberg <si...@gmail.com>]

---------- Forwarded message ----------
From: Sivan Greenberg <si...@omniqueue.com>
Date: Sun, Nov 7, 2010 at 6:58 PM
Subject: Questions about database file sizes.
To: users@couchdb.apache.org


Hi List,

 First I'd like to say that my longly developed couchdb based system
seems to be performing well and serving its purpose.

 However, there's always a small difference between the database files
on the 2 couchdb server nodes we are using, after identical compaction
steps on both db's, the difference remains around 4k, even though the
number of documents is the same.

 This is CouchDB 0.11.0 , 2 nodes, each one does a pull replication
from its peer in continue replication over HTTP.

 Is this normal? Can we sleep at night without fearing the delta would
at some edge case occasion increase to the gigs ? Is it unrealistic to
expect the two files to be of the same size given same compaction
steps were taken and they both replicate pull from each other?

Many thanks,
(CouchDB Rocks!)

-Sivan

Re: Questions about database file sizes.

[Adam Kocoloski <ko...@apache.org>]

Replication won't copy the _local docs between the nodes.  You might have some of those on only one node, e.g. from a replication that only involved one of the two databases.  Regards,

Adam

On Nov 7, 2010, at 3:31 PM, Sivan Greenberg wrote:

> So far, I get the differences even after compacting twice and three times.
> 
> On Sun, Nov 7, 2010 at 10:21 PM, Paul Davis <pa...@gmail.com> wrote:
>> On Sun, Nov 7, 2010 at 12:02 PM, Sivan Greenberg <si...@gmail.com> wrote:
>>> ---------- Forwarded message ----------
>>> From: Sivan Greenberg <si...@omniqueue.com>
>>> Date: Sun, Nov 7, 2010 at 6:58 PM
>>> Subject: Questions about database file sizes.
>>> To: users@couchdb.apache.org
>>> 
>>> 
>>> Hi List,
>>> 
>>>  First I'd like to say that my longly developed couchdb based system
>>> seems to be performing well and serving its purpose.
>>> 
>>>  However, there's always a small difference between the database files
>>> on the 2 couchdb server nodes we are using, after identical compaction
>>> steps on both db's, the difference remains around 4k, even though the
>>> number of documents is the same.
>>> 
>>>  This is CouchDB 0.11.0 , 2 nodes, each one does a pull replication
>>> from its peer in continue replication over HTTP.
>>> 
>>>  Is this normal? Can we sleep at night without fearing the delta would
>>> at some edge case occasion increase to the gigs ? Is it unrealistic to
>>> expect the two files to be of the same size given same compaction
>>> steps were taken and they both replicate pull from each other?
>>> 
>>> Many thanks,
>>> (CouchDB Rocks!)
>>> 
>>> -Sivan
>>> 
>> 
>> You might try compacting twice. If you're not seeing the difference
>> change after continuous writes its probably not a big deal.
>> 
>> Paul
>> 

Re: Questions about database file sizes.

[Sivan Greenberg <si...@omniqueue.com>]

So far, I get the differences even after compacting twice and three times.

On Sun, Nov 7, 2010 at 10:21 PM, Paul Davis <pa...@gmail.com> wrote:
> On Sun, Nov 7, 2010 at 12:02 PM, Sivan Greenberg <si...@gmail.com> wrote:
>> ---------- Forwarded message ----------
>> From: Sivan Greenberg <si...@omniqueue.com>
>> Date: Sun, Nov 7, 2010 at 6:58 PM
>> Subject: Questions about database file sizes.
>> To: users@couchdb.apache.org
>>
>>
>> Hi List,
>>
>>  First I'd like to say that my longly developed couchdb based system
>> seems to be performing well and serving its purpose.
>>
>>  However, there's always a small difference between the database files
>> on the 2 couchdb server nodes we are using, after identical compaction
>> steps on both db's, the difference remains around 4k, even though the
>> number of documents is the same.
>>
>>  This is CouchDB 0.11.0 , 2 nodes, each one does a pull replication
>> from its peer in continue replication over HTTP.
>>
>>  Is this normal? Can we sleep at night without fearing the delta would
>> at some edge case occasion increase to the gigs ? Is it unrealistic to
>> expect the two files to be of the same size given same compaction
>> steps were taken and they both replicate pull from each other?
>>
>> Many thanks,
>> (CouchDB Rocks!)
>>
>> -Sivan
>>
>
> You might try compacting twice. If you're not seeing the difference
> change after continuous writes its probably not a big deal.
>
> Paul
>

Re: Questions about database file sizes.

[Paul Davis <pa...@gmail.com>]

On Sun, Nov 7, 2010 at 12:02 PM, Sivan Greenberg <si...@gmail.com> wrote:
> ---------- Forwarded message ----------
> From: Sivan Greenberg <si...@omniqueue.com>
> Date: Sun, Nov 7, 2010 at 6:58 PM
> Subject: Questions about database file sizes.
> To: users@couchdb.apache.org
>
>
> Hi List,
>
>  First I'd like to say that my longly developed couchdb based system
> seems to be performing well and serving its purpose.
>
>  However, there's always a small difference between the database files
> on the 2 couchdb server nodes we are using, after identical compaction
> steps on both db's, the difference remains around 4k, even though the
> number of documents is the same.
>
>  This is CouchDB 0.11.0 , 2 nodes, each one does a pull replication
> from its peer in continue replication over HTTP.
>
>  Is this normal? Can we sleep at night without fearing the delta would
> at some edge case occasion increase to the gigs ? Is it unrealistic to
> expect the two files to be of the same size given same compaction
> steps were taken and they both replicate pull from each other?
>
> Many thanks,
> (CouchDB Rocks!)
>
> -Sivan
>

You might try compacting twice. If you're not seeing the difference
change after continuous writes its probably not a big deal.

Paul