Apache ZooKeeper

MAHADEV KONAR
 What is ZooKeeper?

A highly available, scalable, distributed

coordination kernel
 Use Cases

» Leader Election
» Group Membership
» Work Queues
» Event Notifications/workflow management
» Configuration Management
» Cluster Management
» Sharding
 What is ZooKeeper again?

 File api without partial reads/writes

 No renames
 Ordered updates and strong persistence
guarantees
 Conditional updates (version)
 Watches for data changes
 Ephemeral znodes
 Generated file names
 Data Model
/
 Hierarchal namespace
apps
 Each znode has data and app1
children servers
 data is read and written
regionserver
in its entirety
master

locks
read-1

users
 ZooKeeper API

String create(path, data, acl, flags)

void delete(path, expectedVersion)

Stat setData(path, data, expectedVersion)

(data, Stat) getData(path, watch)

Stat exists(path, watch)

String[] getChildren(path, watch)

 ZooKeeper Service

ZooKeeper Service
Leader

Server Server Server Server Server

Client Client Client Client Client Client Client

 All servers store a copy of the data (in memory)

 A leader is elected at startup
 Followers service clients, all updates go through leader
 Update responses are sent when a majority of servers have persisted the change
 ZooKeeper and HBase

Master Failover

Region Servers and Master discovery via ZooKeeper

 HBase clients connect to ZooKeeper to find configuration data
 Region Servers and Master failure detecti0n
 Hbase and ZooKeeper as of now!

shutdown
• Master
/ • If more than one master, they fight
root-region-server • Root Region Server
• This znode holds the location of the server
hosting the root of all tables in hbase
rs • rs
• A directory in which there is a znode per
master Hbase region server
• Region Servers register themselves with
ZooKeeper when they come online
• On Region Server failure (detected via ephemeral
znodes and notification via ZooKeeper), the master
splits the edits out per region
 Common Problems/Error Cases

Garbage Collection at the Region Servers

 Causes zookeeper clients to stall
 Session expiry

Low throughput and connection loss

 Mostly due to under provisioned ZooKeeper instances
 Disk and Memory usage

Bad Usage example:

 NameNode, RegionServer, JobTracker, ZooKeeper running on
the same node
 Release 3.3.0, whats in for Hbase?

Allow configuration of session timeout min/max

bounds
 HBase needs large session timeouts
Improved logging information to detect issues
Improved debugging tools
Improved documentation
Improved performance and robustness
Queue implementation available
 Upcoming 3.4 release

No Connectionloss
Use Netty - allow encryption
Testing
 Mockito
More of backwards compatibility testing
 More ZooKeeper in Hbase?

Table Schema and state in ZooKeeper

 read only, online
Region Server state transitions via ZooKeeper
Store region assignment in ZooKeeper for each
Region Server
http://wiki.apache.org/hadoop/ZooKeeper/
HBaseUseCases
Questions?