You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@couchdb.apache.org by Franck Eyraud <fr...@yrnm.net> on 2013/08/29 20:31:41 UTC
Lots of requests with continuous replications (every 5 seconds)
Hello list,
I didn't find the answer to the following questions using Google :
Why does Couchdb send a GET request to the database every 5 seconds
while replicating continuously ? Doesn't the replicator use the _change
API (with heartbeat functionality) anyway ? And if there is a need to
poll aserver alive, 5 seconds seem very often to me.
Plus, it seems it adds some fee issues in some cases :
http://stackoverflow.com/questions/18254772/hanging-replications-on-cloudant
I don't have the issue above yet, but in the case of an application
where clients can replicate so that they can run offline, this would
result in a loot of requests. Is there a way to avoid this ? To make it
less often ?
Thank you for your help,
Franck Eyraud
Re: Lots of requests with continuous replications (every 5 seconds)
Posted by Robert Newson <rn...@apache.org>.
It certainly is silly, yes.
B.
On 29 August 2013 19:38, Adam Kocoloski <ko...@apache.org> wrote:
> It's really a bit silly, though -- as far as I can tell the GET requests against the source DB are only used to keep the source_seq information in _active_tasks current.
>
> Adam
>
> On Aug 29, 2013, at 2:36 PM, Robert Newson <rn...@apache.org> wrote:
>
>> Hi,
>>
>> The checkpoint interval is now configurable on Cloudant (as of a few
>> days ago) as a consequence of either your report or another user. Just
>> add "checkpoint_interval": <number> to your _replicate request. If
>> your replication is started by CouchDB then this option will not
>> currently work.
>>
>> B.
>>
>>
>>
>> On 29 August 2013 19:31, Franck Eyraud <fr...@yrnm.net> wrote:
>>> Hello list,
>>>
>>> I didn't find the answer to the following questions using Google :
>>>
>>> Why does Couchdb send a GET request to the database every 5 seconds while
>>> replicating continuously ? Doesn't the replicator use the _change API (with
>>> heartbeat functionality) anyway ? And if there is a need to poll aserver
>>> alive, 5 seconds seem very often to me.
>>>
>>> Plus, it seems it adds some fee issues in some cases :
>>> http://stackoverflow.com/questions/18254772/hanging-replications-on-cloudant
>>>
>>> I don't have the issue above yet, but in the case of an application where
>>> clients can replicate so that they can run offline, this would result in a
>>> loot of requests. Is there a way to avoid this ? To make it less often ?
>>>
>>> Thank you for your help,
>>>
>>> Franck Eyraud
>>>
>>>
>>>
>
Re: Lots of requests with continuous replications (every 5 seconds)
Posted by Adam Kocoloski <ko...@apache.org>.
It's really a bit silly, though -- as far as I can tell the GET requests against the source DB are only used to keep the source_seq information in _active_tasks current.
Adam
On Aug 29, 2013, at 2:36 PM, Robert Newson <rn...@apache.org> wrote:
> Hi,
>
> The checkpoint interval is now configurable on Cloudant (as of a few
> days ago) as a consequence of either your report or another user. Just
> add "checkpoint_interval": <number> to your _replicate request. If
> your replication is started by CouchDB then this option will not
> currently work.
>
> B.
>
>
>
> On 29 August 2013 19:31, Franck Eyraud <fr...@yrnm.net> wrote:
>> Hello list,
>>
>> I didn't find the answer to the following questions using Google :
>>
>> Why does Couchdb send a GET request to the database every 5 seconds while
>> replicating continuously ? Doesn't the replicator use the _change API (with
>> heartbeat functionality) anyway ? And if there is a need to poll aserver
>> alive, 5 seconds seem very often to me.
>>
>> Plus, it seems it adds some fee issues in some cases :
>> http://stackoverflow.com/questions/18254772/hanging-replications-on-cloudant
>>
>> I don't have the issue above yet, but in the case of an application where
>> clients can replicate so that they can run offline, this would result in a
>> loot of requests. Is there a way to avoid this ? To make it less often ?
>>
>> Thank you for your help,
>>
>> Franck Eyraud
>>
>>
>>
Re: Lots of requests with continuous replications (every 5 seconds)
Posted by Franck Eyraud <fr...@yrnm.net>.
In fact, I am not using cloudant, just a self hosted couchdb server, and
a standalone workstation that starts its own replication, always with
couchdb 1.2, or 1.3.
I just added the link because it is the only one close to my problem I
found on the web.
So this checkpoint_interval is not possible to change wiht plain couchdb
server ?
And what this checkpoint is for ?
Thanks for your help,
Franck
PS: I have to check but one of the workstation was running Couchdb 1.3.1
on a macintosh, and the requests where every 30 seconds, and with a
simple HEAD, not a GET.
On 08/29/2013 08:36 PM, Robert Newson wrote:
> Hi,
>
> The checkpoint interval is now configurable on Cloudant (as of a few
> days ago) as a consequence of either your report or another user. Just
> add "checkpoint_interval": <number> to your _replicate request. If
> your replication is started by CouchDB then this option will not
> currently work.
>
> B.
>
>
>
> On 29 August 2013 19:31, Franck Eyraud <fr...@yrnm.net> wrote:
>> Hello list,
>>
>> I didn't find the answer to the following questions using Google :
>>
>> Why does Couchdb send a GET request to the database every 5 seconds while
>> replicating continuously ? Doesn't the replicator use the _change API (with
>> heartbeat functionality) anyway ? And if there is a need to poll aserver
>> alive, 5 seconds seem very often to me.
>>
>> Plus, it seems it adds some fee issues in some cases :
>> http://stackoverflow.com/questions/18254772/hanging-replications-on-cloudant
>>
>> I don't have the issue above yet, but in the case of an application where
>> clients can replicate so that they can run offline, this would result in a
>> loot of requests. Is there a way to avoid this ? To make it less often ?
>>
>> Thank you for your help,
>>
>> Franck Eyraud
>>
>>
>>
Re: Lots of requests with continuous replications (every 5 seconds)
Posted by Robert Newson <rn...@apache.org>.
Hi,
The checkpoint interval is now configurable on Cloudant (as of a few
days ago) as a consequence of either your report or another user. Just
add "checkpoint_interval": <number> to your _replicate request. If
your replication is started by CouchDB then this option will not
currently work.
B.
On 29 August 2013 19:31, Franck Eyraud <fr...@yrnm.net> wrote:
> Hello list,
>
> I didn't find the answer to the following questions using Google :
>
> Why does Couchdb send a GET request to the database every 5 seconds while
> replicating continuously ? Doesn't the replicator use the _change API (with
> heartbeat functionality) anyway ? And if there is a need to poll aserver
> alive, 5 seconds seem very often to me.
>
> Plus, it seems it adds some fee issues in some cases :
> http://stackoverflow.com/questions/18254772/hanging-replications-on-cloudant
>
> I don't have the issue above yet, but in the case of an application where
> clients can replicate so that they can run offline, this would result in a
> loot of requests. Is there a way to avoid this ? To make it less often ?
>
> Thank you for your help,
>
> Franck Eyraud
>
>
>