[MarkLogic Dev General] cpf question

[Mike Sokolov]

Thanks for your quick response, Mary.

In the course of digging a bit further, I found I had neglected to 
attach my pipeline to the domain.  I think things are moving along OK now.

-Mike


Mary Holstege wrote:
> On Mon, 21 Jul 2008 07:46:00 -0700, Mike Sokolov 
> <sokolov at ifactory.com> wrote:
>
>> I have set up a very simple content processing pipeline; basically there
>> is a single action to be taken whenever any document is updated or
>> created.  I've gotten this to work well enough when starting with an
>> empty database.  However now I am in a state where I can't seem to get
>> my actions to be triggered by updating the already-existing documents.
>> I have two questions:
>>
>> 1) If you load all your documents, and then install cpf, load your
>> actions and set up a pipeline, will this trigger processing?  Is there
>> some way to get documents to be processed at that point?
>
> CPF responds to events, so to reprocess existing content, you need
> to touch each of those documents in order to get CPF to notice them.
>
>> 2) What does "TaskServer: state is already successor so not updating for
>> <uri>" mean?  It sounds as if TaskServer is saying: hey I processed this
>> document already, so (even though you just updated it), I'm not going to
>> process it again.  How can I set things up so updates work?
>
> This is benign.  It was put in to trace cases where multiple actions
> were interfering with each other, but it turns out the way it is testing
> for that case was flawed and it gets reported more than it should.
>
>> OK, third question: what is the recommended "best practice" for updating
>> all the documents, in order to get the maximum loading throughput.  This
>> is in a development system, so I can handle some downtime. Do you wipe
>> the database first and then reload?
>
> That is maybe the simplest, but all you need to do to get CPF to pay 
> attention
> is to touch each document; if they have multiple fragments, that could be
> cheaper than a full reload.
>
>> I am attaching a copy of my simple pipeline in case it has some 
>> relevance...
>
> What other pipelines do you have attached to the domain?  You will
> need the status handling pipeline.
>
> //Mary
> _______________________________________________
> General mailing list
> General at developer.marklogic.com
> http://xqzone.com/mailman/listinfo/general