Hi, I understand that tail can be applied to multiple files at once but that will combine all output together and my intension is to keep log data from different files separate from each other.
Another way is to treat each file to monitor as distinct sources and then write output to distinct channels (file / memory) and shipping them to different sinks. agent.sources = s1 s2 ... sn agent.channels = fc1 fc2 ... fcn agent.sinks = sk1 sk2 ... sn agent.sources.s1.exec = tail -F -n 0 /var/log/myapp1.log agent.sources.s2.exec = tail -F -n 0 /var/log/myapp2.log . . agent.sources.sn.exec = tail -F -n 0 /var/log/myappn.log Now s1 ---> fc1 ---> sk1 s2 ---> fc2 ---> sk2 s3 ---> fc3 ---> sk3 .. .. etc. I am looking for a generic way of achieving the same as explained above if possible. Thanks, - dib On Thu, Mar 21, 2013 at 2:51 PM, Dibyajyoti Ghosh <[email protected] > wrote: > Hi flume-dev, > > In flume v1.3.0 bundled with cdh-4.2.0 I am aware of exec source where one > can do 'tail -f' for a single mutable file. > > Spool-dir is available but files have to be uniquely named and immutable. > > Is there some way I can tail a directory much like multi tail / tailDir > option in the current flume version? > > thanks a lot, > - dib > On Thu, Mar 21, 2013 at 2:51 PM, Dibyajyoti Ghosh <[email protected] > wrote: > Hi flume-dev, > > In flume v1.3.0 bundled with cdh-4.2.0 I am aware of exec source where one > can do 'tail -f' for a single mutable file. > > Spool-dir is available but files have to be uniquely named and immutable. > > Is there some way I can tail a directory much like multi tail / tailDir > option in the current flume version? > > thanks a lot, > - dib >
