[
https://issues.apache.org/jira/browse/NIFI-473?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15670868#comment-15670868
]
Mark Payne commented on NIFI-473:
---------------------------------
[~rickysaltzer] This is a fairly old ticket, but I wanted to revisit, as in 1.1
we now have the ability to separate the notions of CRON scheduling and Primary
Node. In the case that you are describing here, I would recommend that perhaps
the proper solution is to use CRON scheduling, rather than timer-driven
scheduling. Do you think that's a reasonable approach, or do you feel like
something like this is still necessary?
Thanks
-Mark
> Add rolling restart / refresh
> ------------------------------
>
> Key: NIFI-473
> URL: https://issues.apache.org/jira/browse/NIFI-473
> Project: Apache NiFi
> Issue Type: New Feature
> Reporter: Ricky Saltzer
>
> For long running instances and clusters, it's sometimes required to switch
> out custom NARs for new or updated processors. Currently, when you restart a
> NiFI instance, processors that are on a timer driven schedule (e.g. 1 day, 1
> week) are executed, regardless of their schedule. This can be problematic if
> you're dealing with datasets that shouldn't be pulled out of schedule.
> It would be nice to have a rolling restart, or a refresh ability to load in
> new NARs without re-executing all of the processors on the system. We could
> take this one step further and have all processor schedules persisted
> somewhere, and is read on start up.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
