>> Also, as far as BookKeeper concerned -- I'd live to understand >> what's involved in using it as a HA solution for HDFS and whether >> there's anybody interested in pursuing this direction. As Hadoop-2 has a chance to plug in the Journal implementations, BookKeeperJournalManager is the implementation for that. We are maintaining this in Hadoop-2 code and all the improvements for the integration mudule is getting tracked with a JIRA HDFS-3399. In HA case, we will configure this bk url as shared edits path. Here shared storage server would be a BookKeeper servers. Here is the usage related docs : http://hadoop.apache.org/docs/current/hadoop-yarn/hadoop-yarn-site/HDFSHighAvailability.html#BookKeeper_as_a_Shared_storage_EXPERIMENTAL
Regards, Uma ________________________________________ From: Flavio Junqueira [fpjunque...@yahoo.com] Sent: Tuesday, January 08, 2013 4:25 PM To: bookkeeper-dev@zookeeper.apache.org Cc: Roman Shaposhnik; Uma Maheswara Rao G; Sijie Guo; Flavio Junqueira Subject: Re: [ANNOUNCE] Apache Bigtop 0.5.0 release As for use cases, we have a poweredby page listing the ones we know of: https://cwiki.apache.org/confluence/display/BOOKKEEPER/PoweredBy -Flavio On Jan 8, 2013, at 9:36 AM, Ivan Kelly <iv...@yahoo-inc.com> wrote: > I think to start we could create bigtop components for > bookkeeper-server and bkjournal manager, as these fix most directly > into the hadoop eco-system. > > I haven't looked much into what is involved in packaging for bigtop, > but I can start looking once the BK 4.2.0 release is out (should be > thursday or friday this week). This is the release we would aim to put > in our initial bigtop release. > > + cc: bookkeeper-dev@ > > -Ivan > > > On Tue, Jan 08, 2013 at 01:08:52AM +0000, Roman Shaposhnik wrote: >> On Mon, Jan 7, 2013 at 5:06 AM, Uma Maheswara Rao G >> <mahesw...@huawei.com> wrote: >>> Wish you a very happy new Year >>> Thanks a lot, Roman for accepting to include. :-) >>> Thanks a lot to Ivan and Sijie for offering the help to Bigtop. I can also >>> support, if I get a chance to do that. >> >> Perfect! For the next steps I think we've got to identify a few >> existing use cases (or better yet active users!) for the BK and >> Hedwig. It is much easier to work from an already existing >> practices of deployment, etc. >> >> Also, as far as BookKeeper concerned -- I'd live to understand >> what's involved in using it as a HA solution for HDFS and whether >> there's anybody interested in pursuing this direction. >> >> Finally, perhaps this conversation belongs to @zookeeper.a.o & >> @bigtop.a.o >> >> Thanks, >> Roman.
