Thanks Ravi, will look into possible alternatives to look for adding dynamic columns while ingesting or open a jira request.
Another question regarding the Flume ingestion. I got multiple clusters with Flume and Phoenix running on different clusters. Are there any libs that need to be in Flume's classpath along with the zookeeperQuorum to pointing to the appropriate cluster? Thanks! On Thu, Jun 11, 2015 at 12:58 PM, Ravi Kiran <[email protected]> wrote: > Hi Buntu, > > Apparently, the necessary classes related to Flume client are already > part of the phoenix client. Hence, you wouldn't need to build a Flume > client artifact additionally. > In the current implementation, the UPSERT query is based on an existing > table only . However, if you would like to have support for dynamic columns > , please do raise a JIRA . > We can definitely add features based on your feedback. > > Regards > Ravi > > > > On Wed, Jun 10, 2015 at 10:15 PM, Buntu Dev <[email protected]> wrote: > >> I plan on using Apache Phoenix on CDH and installed the parcel >> (4.3.0-clabs-phoenix-1.0.0-all). Does the Flume client still needs to be >> built for the client jar as mentioned in the install and setup steps here: >> http://phoenix.apache.org/flume.html >> >> I'm also looking for some examples of ingesting events to Phoenix via >> Flume. Any pointers on creating a table and dynamically adding column >> families and columns during ingestion will be really helpful. >> >> >> Thanks! >> > >
