There are 2 approached NiFi processors follow today.

E.g. a new TailFile saves it in a local state file (relative to NiFi root, in 
the ./conf/ directory).

Others leverage a DistributedCache service (it's actually more like a 
RemoteCache, not a distributed/replicated), which they declare as a requirement 
in the properties. This also has a benefit of being able to share the state 
between nodes and, optionally, persist the cache to the disk to be able to 
re-hydrate it on restart.

Andrew



On 11/13/15, 10:09 AM, "Sean Busbey" <[email protected]> wrote:

>Sam,
>
>For now I would store this information in HDFS. Alternatively, if you have
>it ZooKeeper would also work.
>
>On Thu, Nov 12, 2015 at 10:45 PM, Sam Kamau <[email protected]> wrote:
>
>> I am working on a processor that loads data to HDFS for an RDBMS using
>> ExecuteSQL processor. The data has a last update time that I would like to
>> use to load records that have changed since the last load.
>>
>> How best can I go about saving the last execution time so I can use it as a
>> filter in the query.
>>
>> Thanks
>>
>
>
>
>-- 
>Sean

Reply via email to