PutElasticsearch currently requires you have the id for the document
in a flowfile attribute. It does not yet support auto-increment, but
you can emulate that with an UpdateAttribute processor before
PutElasticsearch, set something like "es.docid" to ${nextInt()}, then
in PutElasticsearch the Identifier Attribute would be es.docid

Regards,
Matt

On Sat, Jun 25, 2016 at 12:55 AM, john lozada <[email protected]> wrote:
> Bryan,
>
> thanks for the information.  In the putelasticsearch processor, what
> exactly do I put in
> *Identifier Attribute ? *
>
>
>
>
>
>
>
>
> Hello,
>
> For Syslog it is relatively straight forward if you have ListenSyslog set
> to parse messages (or use the ParseSyslog processor). That will produce
> attributes on each flow file which could easily be converted to JSON with
> the AttributesToJson processor.
>
> This blog post shows how to do this:
> https://blogs.apache.org/nifi/entry/storing_syslog_events_in_hbase
>
> In general, the best case is if the logs can be produced in json format at
> the source, which is usually easy to do with applications that use log4j or
> log back, but of course a lot of times the producer of the logs can't be
> changed.
>
> -Bryan
>
> On Friday, June 24, 2016, john lozada <[email protected]> wrote:
>
>> helo,
>>
>> Is there a way to make syslogs/ or any application logs into .json format?
>> I just started to use elasticsearch and from my understanding the logs
> have
>> to be in .json format. We are using logstash for the moment but, I would
>> prefer to be able to replace it with nifi.
>>

Reply via email to