incubator-couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andreas Kemkes <a5s...@yahoo.com>
Subject Re: Replications and checkpoints (again)
Date Wed, 22 Aug 2012 22:39:40 GMT
Yes, the replication replicates - there are PUTs and POSTs in the log that could only come
from the replication - no other process updates the replica.


doc_count is kept in sync on both source and target.

The docs are big (attachments), so 1K docs would be quite a bit of data.

source_seq is growing, checkpointed_source_seq still at the same number.  Could the difference
in the disk_format_version be the reason?

Is there some documentation about how replications and checkpoints relate to each other?

Thanks for the info on the null value for data_size.

-- Andreas


________________________________
 From: Paul Davis <paul.joseph.davis@gmail.com>
To: user@couchdb.apache.org; Andreas Kemkes <a5sk4s@yahoo.com> 
Sent: Wednesday, August 22, 2012 3:14 PM
Subject: Re: Replications and checkpoints (again)
 
Did the replication actually replicate anything or were they up to
date when started? I'm not sure on the exact algorithm continuous
replication uses but I wouldn't be surprised if its just every N docs.
The example you have is less than 1K docs which wouldn't surprise me
as the threshold.

data_size being null probably means you haven't compacted that db
since upgrade CouchDB.

On Mon, Aug 20, 2012 at 1:00 PM, Andreas Kemkes <a5sk4s@yahoo.com> wrote:
> I'm still wondering about the behavior of a continuous replication and the status feedback
as far as checkpoints are concerned.
>
> Both Futon and the _active_tasks api call indicate that the check-pointed source sequence
is lagging behind:
>
> [{"pid":"<0.12494.0>",
> "checkpointed_source_seq":990974,
> "continuous":true,...,
> "replication_id":"06058d20f039a2e8d990577ba65ae35e+continuous",...,
> "source_seq":991539,...,
> "type":"replication",
> "updated_on":1345480111}]
>
> Yet both databases show the same amount of documents:
>
>
> {"db_name":"source",
> "doc_count":157713,
> "doc_del_count":9021,
> "update_seq":991539,
> "purge_seq":0,
> "compact_running":false,
> "disk_size":201812152456,
> "data_size":null,
> "instance_start_time":"1343849350087635",
> "disk_format_version":5,
> "committed_update_seq":991539}
>
> {"db_name":"target",
> "doc_count":157713,
> "doc_del_count":9021
> ,"update_seq":173173,
> "purge_seq":0,
> "compact_running":false,
> "disk_size":183233859719,
> "data_size":176762791873,
> "instance_start_time":"1343849350176012",
> "disk_format_version":6,
> "committed_update_seq":173173}
>
> What is the expected behavior?
>
> Why would there be no update to a checkpoint during the continuous replication?
>
> Also, what is the reason for the data_size being null in the source database?
>
> Thanks in advance.
>
> -- Andreas
Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message