Hi everyone, I was discussing open sourcing some of the work we have done in bookkeeper at SFDC with @Venkateswara Rao Jujjuri <jujj...@gmail.com>. I'm starting this thread as it would be interesting to gauge interest of the community and get some feedback as some of this work addresses gaps that exist in the community and would be beneficial for everyone.
To start with we have what we call the *ZK metadata reconstruction *tool. It addresses the story around backup-recovery / disaster recovery for BK. The use case behind this tool is that if the data in the metadata server is lost, we lose the information about ledgers and ensembles. This tool connects to all the bookie nodes and prepares the metadata about the data stored in the bookies using the information available in the bookies - which was previously stored in ZK. The tool then reconstructs this metadata and writes to it ZK returning it to the state it was before it lost all the information. We understand that the users of BK are varied - where some users have small clusters and others have their own solutions to backup and recovery. However it would be good to see some support / interest from the devs outside of SFDC to champion this effort for the Apache community. What do you folks think? Regards, Anup -- Anup Ghatage www.ghatage.com
