It's possible you could use the exec source and tail -F but personally I would use either the Flume 1.3 RC (until it's released).
On Thu, Nov 29, 2012 at 11:52 AM, Robert Slifka <[email protected]>wrote: > Hi Brock, > > Thanks for the heads up, wasn't aware that was incoming. > > If you don't mind me asking, what's the current recommended approach with > 1.2? > > Rob > > > On Thu, Nov 29, 2012 at 9:19 AM, Brock Noland <[email protected]> wrote: > >> I would use the Spool Directory source in 1.3 to upload the logs files >> once apache rolls the log. >> >> >> On Wed, Nov 28, 2012 at 6:33 PM, Robert Slifka >> <[email protected]>wrote: >> >>> Hi all, >>> >>> Simple HTTP access.log aggregation case here. Front ends running Apache >>> needing to aggregate/sink all access.log data to S3. >>> >>> Right now we're on OG 0.9.3: >>> >>> nginx => access.log => tailSource => Collector => S3 >>> >>> >>> For 1.2, I was under the impression we would simplify and move to: >>> >>> Apache => Syslog => Agent…Agent => S3. >>> >>> >>> However it looks like Apache can only Syslog the error.log. Am I >>> missing something? >>> >>> What's the recommended 'standard' setup? >>> >>> Thanks guys, >>> >>> Rob >>> >>> >> >> >> -- >> Apache MRUnit - Unit testing MapReduce - >> http://incubator.apache.org/mrunit/ >> > > -- Apache MRUnit - Unit testing MapReduce - http://incubator.apache.org/mrunit/
