[GitHub] [accumulo] jzgithub1 opened a new pull request #1382: Fixes #1225 - Place all table configurations under single Zookeeper path.

[GitBox <gi...@apache.org>]

jzgithub1 opened a new pull request #1382: Fixes #1225 - Place all table configurations under single Zookeeper path.
URL: https://github.com/apache/accumulo/pull/1382
 
 
   This pull request adds a new Zookeeper path to put the table configurations for all tables under a single Znode.  This path is located at  /accumulo/{instance-id}/table_configs  in Zookeeper.  Many of the other table configurations will be still located under  /accumulo/{instance-id}/tables to maintain the current functional operation of the code with the present ZooCache class which has not been modified at all.  The relocated configurations are ones like these:  table.iterator.majc.vers, table.constraint.1, table.iterator.scan.vers.opt.maxVersions, table.iterator.minc.vers, table.iterator.majc.vers.opt.maxVersions, table.iterator.minc.vers.opt.maxVersions, table.iterator.scan.vers.  
   
   Those configuration values were being cleared from the ZooCache during  add/remove/clone table operations. This necessitated calling Zookeeper again for all the configurations of all the tables.  For large systems that constantly added and removed tables this caused ZooCache.get to created new Watchers on configurations that ZooKeeper already had a watch on.  On systems with many TServers with many tables this must have put a larger burden on Zookeeper than we would see in smaller systems.
   I think that this pull request reduces the number of Zookeeper Watchers that get created in the ZooCache.get function.  I have observed this in trace debugging.  
   
   I have run the ingest testing in the Accumulo-Testing project on this code.  Also I have verified table creation, deletion and cloning in the accumulo command line interface.  I verified the creation of the table configurations at their new locations inside of the ZooKeeper command line interface. 

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services