not sure if its the same for embedded agent but our standalone flume daemons watch their config for changes and do a graceful reload when changes are detected (without dropping events in the channel)
we use ansible to push changes to our sink config on upstream tiers when we add nodes to our downstream flume aggregate tiers should be easy to test by touching your flume config and seeing if its re-initializes and spits out info to log with startup info -- Iain Wright This email message is confidential, intended only for the recipient(s) named above and may contain information that is privileged, exempt from disclosure under applicable law. If you are not the intended recipient, do not disclose or disseminate the message to anyone except the intended recipient. If you have received this message in error, or are not the named recipient(s), please immediately notify the sender by return email, and delete all copies of this message. On Tue, Oct 28, 2014 at 4:54 PM, Pavel Zalunin <[email protected]> wrote: > Hi, > > We have embedded agent, zookeeper is huge overhead. > > Pavel. > > 2014-10-28 20:11 GMT+03:00 Paweł Róg <[email protected]>: > >> Hi, >> And as far as I know flume can be configured by zookeeper :D >> >> https://issues.apache.org/jira/browse/FLUME-1491 >> >> -- >> Pawel Rog >> >> 2014-10-28 17:02 GMT+01:00 Otis Gospodnetic <[email protected]>: >> >>> As far as I know, Flume is configured only through the config and I >>> *think* Flume automatically reloads the config when it detects the config >>> changed - at least that is what Ashish wrote here: >>> http://search-hadoop.com/m/DEeB4w0kHg1 >>> >>> Otis >>> -- >>> Monitoring * Alerting * Anomaly Detection * Centralized Log Management >>> Solr & Elasticsearch Support * http://sematext.com/ >>> >>> >>> On Mon, Oct 27, 2014 at 7:42 PM, Pavel Zalunin <[email protected]> >>> wrote: >>> >>>> Hi, >>>> >>>> We have embedded flume agent and we need to update some parameters >>>> dynamically after sink/source runners are started. >>>> >>>> For example, having sink which sends events to some url. At some point >>>> we need to change this url (add request parameter for example). >>>> >>>> Straightforward way is to add method which will accept changes and >>>> update internal state, but it looks like a 'hack', because all flume sinks, >>>> that I've seen configured just once. >>>> >>>> Maybe there are another, 'flume-way' to do such things? >>>> >>>> Is restarting sink (I think restart other related things also needed) >>>> only one option? >>>> >>>> Thanks, >>>> Pavel. >>>> >>> >>> >> >
