incubator-couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Robert Newson <rnew...@apache.org>
Subject Re: _replicator database and proxies
Date Wed, 24 Apr 2013 15:31:48 GMT
I think that depends on the setting of max_replication_retry_count in config.

On 24 April 2013 11:27, Steven Barlow <stemail23@gmail.com> wrote:
> That did occur to me too. Perhaps it's a viable solution, but it
> doesn't seem very elegant.
>
> Also (and I haven't investigated thoroughly, so I don't know the
> answer) I'm not totally clear on whether the failed replications
> (those in error state due to invalid proxy config for the current
> situation) would be expected to resume automatically when their proxy
> settings became valid again?
>
> On 25/04/2013, at 12:16 AM, Robert Newson <rnewson@apache.org> wrote:
>
>> Hm, why not have a replication document for each of those two
>> situations? One with proxy, one without?
>>
>> B.
>>
>> On 24 April 2013 10:11, Steven Barlow <stemail23@gmail.com> wrote:
>>> Thanks for the comments.
>>>
>>> I appreciate that each replication document is independent. I guess
>>> I'm questioning whether the proxy configuration is typically an
>>> attribute of the individual replication entity, rather than being a
>>> "system-wide" connection configuration. I'm struggling to envisage a
>>> scenario whereby different replication documents might have different
>>> proxy settings?
>>>
>>> To support that use case though, I guess, why not retain a replication
>>> document level setting as an override, but introduce a global setting
>>> for what, at least in my opinion, is the more typical use case?
>>>
>>> In answer to the question, "why is my configuration changing so
>>> rapidly?" - My proxy settings change several times every day when my
>>> laptop travels to and from my home network (no proxy), to my office
>>> network (which requires a proxy for Internet access).
>>>
>>>
>>>
>>> On 24/04/2013, at 11:42 PM, Robert Newson <rnewson@apache.org> wrote:
>>>
>>>> Q: "Why is the proxy configuration part of the replication document at all?
>>>>
>>>> A: Because each replication document is an independent entity.
>>>>
>>>> Making it global works for your case but not the general case. Why is
>>>> your proxy configuration changing so rapidly?
>>>>
>>>> B.
>>>>
>>>> On 24 April 2013 08:31, Steven <stemail23@gmail.com> wrote:
>>>>> OK, please excuse the bump, but really? No takers on this one?
>>>>>
>>>>> I'm surprised if I'm the only person encountering this situation. Is
nobody
>>>>> else working on a system with replication to laptop devices that use
>>>>> multiple connections?
>>>>>
>>>>> Here's my biggest confusion: Why is the proxy configuration part of the
>>>>> replication document at all? Surely a global proxy setting would make
way
>>>>> more sense, otherwise, the only way I can see around this issue is to
tear
>>>>> down the entire replication database document set, and rebuild it when
I
>>>>> detect a change in proxy configuration. That seems like madness!
>>>>>
>>>>> Any comments appreciated.
>>>>>
>>>>> Cheers.
>>>>>
>>>>>
>>>>> On Fri, Apr 12, 2013 at 1:14 PM, Steven <stemail23@gmail.com> wrote:
>>>>>
>>>>>> Question about using the _replicator database to set up persistent
>>>>>> replications from a machine that uses various different connections,
>>>>>> sometimes behind a proxy, other times not.
>>>>>>
>>>>>> Question is: Given that the proxy server information is part of the
>>>>>> replication document, the replication breaks (returns an error) when
the
>>>>>> connection changes to use a different, or no, proxy. How am I supposed
to
>>>>>> handle this situation?
>>>>>>
>>>>>> Also, an observation: when the replication document contains "proxy":
null
>>>>>> (in V1.3), the replication fails, this caught me out for a while,
the
>>>>>> replication document proxy parameter must be entirely absent to specify
no
>>>>>> proxy.
>>>>>>

Mime
View raw message