We definitely saw a change, when setting the pipeline size to 1 (disabling it). 
Reading the logs it looks like at least this value is not ignored.

The initial problem to our error was already solved by adam on IRC yesterday 
with this patch http://friendpaste.com/5lLJ0ORh3KPCm1xwwY8Tye


On 21.12.2010, at 03:04, Randall Leeds wrote:

> This is because pull replication ignores these settings. Sorry I keep
> forgetting to put it in JIRA, it's a two line fix.
> 
> Ill do it when I'm home tonight if no one beats me to it.
> 
> couch_rep_reader:init if I remember correctly. It's passing the constants
> instead of the .ini values.
> 
> (Sent while mobile, or i'd just upload a patch right now)
> On Dec 20, 2010 2:14 AM, "Sebastian Cohnen" <[email protected]>
> wrote:
>> We can reproduce the error with session and pipeline values up to 100. we
> always get the same error.
>> 
>> On 17.12.2010, at 19:26, Filipe David Manana wrote:
>> 
>>> On Fri, Dec 17, 2010 at 6:03 PM, Sebastian Cohnen
>>> <[email protected]> wrote:
>>>> I'll try that, thanks. But isn't the reported error a bit odd? why is
> couch trying to JSON encode a erlang pid? at least it looks like it is
> trying to do so...
>>> 
>>> It's because there's a sort of try catch handler which tries to json
>>> encode every exception/error. I believe this should be fixed (or
>>> minimized) in the 1.0.x branch.
>>> 
>>>> 
>>>> 
>>>> On 17.12.2010, at 17:39, Filipe David Manana wrote:
>>>> 
>>>>> Hi Sesbatian,
>>>>> 
>>>>> Try increasing the HTTP sessions and pipeline values in the .ini:
>>>>> 
>>>>> 
> https://github.com/fdmanana/couchdb/blob/trunk/etc/couchdb/default.ini.tpl.in#L138
>>>>> 
>>>>> regards
>>>>> 
>>>>> On Fri, Dec 17, 2010 at 4:33 PM, Sebastian Cohnen
>>>>> <[email protected]> wrote:
>>>>>> Hi,
>>>>>> 
>>>>>> the setup I've described in COUCHDB-968 is causing trouble (again).
> Pull replication is reproducable dying with
> {exit,{json_encode,{bad_term,<0.133.0>}}}. I've uploaded the log [1] from
> which I've changed credentials, db and host name. The log is also truncated
> since it was >40k lines and it looks like documents were included too.
>>>>>> 
>>>>>> Thanks in advance!
>>>>>> 
>>>>>> Sebastian
>>>>>> 
>>>>>> 
>>>>>> [1] http://friendpaste.com/4shMkdhCwmZQY9ITbSGu5T
>>>>>> 
>>>>>> 
>>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> --
>>>>> Filipe David Manana,
>>>>> [email protected], [email protected]
>>>>> 
>>>>> "Reasonable men adapt themselves to the world.
>>>>> Unreasonable men adapt the world to themselves.
>>>>> That's why all progress depends on unreasonable men."
>>>> 
>>>> 
>>> 
>>> 
>>> 
>>> --
>>> Filipe David Manana,
>>> [email protected], [email protected]
>>> 
>>> "Reasonable men adapt themselves to the world.
>>> Unreasonable men adapt the world to themselves.
>>> That's why all progress depends on unreasonable men."
>> 

Reply via email to