Skip to content

Latest commit

 

History

History
239 lines (231 loc) · 6.07 KB

USAGE.md

File metadata and controls

239 lines (231 loc) · 6.07 KB
Raw

How To Use

Node setting

Normally, use Flare-tools to maintain flare by command line.

First of all, you should access index server (flarei) by telnet.

$ telnet [index server] [index server port]

After that, you can configure node servers from index server.
You can check node states like this:

stats nodes

You can set role to each node server like this:

node role [node server name] [node server port] [role=(master|slave|proxy)] [balance] ([partition])

You can set status of each node server like this:

node state [index server] [index server port] [state=(down|active)]

You can remove node server from index server like this:

node remove [index server] [index server port]

Store and get keys

Connect proxy, or master node by telnet.

$ telnet [proxy or master node server] [server port]

You can store keys by set.

set [key name] [flag] [expiration time] [size of value]
[value]

and you can get keys by get.

get [key name]

Example

For example, we introduce how to build flare cluster as the figure below indicates:
sample_image

  • two master node
    • Only master exec write request
  • two slave node
    • if master node down, slave became master automatically
    • slave exec read request
  • one proxy node
    • proxy divide read request in half for master and slave
  • one index server
    • monitor all node
  • two partition
    • each partition can install to another location (like region in AWS)

Create index server

Create config file /etc/flarei.conf as below.

# data directory
data-dir = /home/flarei
# syslog facility
log-facility = local0
# max-connections to accept
max-connection = 256
# node down threshold
monitor-threshold = 3
# node monitoring interval (index server sends "ping" w/ this interval)
monitor-interval = 1
# server name of myself (IP address is preferred)
server-name = 10.0.0.1
# server port of myself
server-port = 12120
# max thread pool size
thread-pool-size = 8

Create data-dir.

$ sudo mkdir /home/flarei

and start index server.

$ sudo /usr/local/bin/flarei -f /etc/flarei.conf --daemonize -p /var/run/flarei.pid

Create master, slave and proxy node

Now, create two master, two slave and one proxy node.
Create config file /etc/flared.conf at each server as below.

# data directory
data-dir = /home/flared
# name and port of index server (sync this to "server-name" of flarei.conf at index server)
index-servers = 10.0.0.1:12120
# syslog facility
log-facility = local1
# max-connections to accept
max-connection = 256
# number of lock slots to access records
mutex-slot = 32
# number of proxy request concurrency
proxy-concurrency = 2
# server name of myself (IP address is preferred)
server-name = 10.0.0.5
# server port of myself
server-port = 12121
# storage type (currently only "tch" is available)
storage-type = tch
# storage options
#storage-ap = 4
#storage-bucket-size = 16777216
#storage-compress =
#storage-large = true
# max number of thread pool
thread-pool-size = 16
# stack size of each thread (kb)
#stack-size = 128

In server-name setting, we change it at each server in accordance with the above figure.
Create data-dir at each server.

$ sudo mkdir /home/flared

Start flared at each server.

$ sudo /usr/local/bin/flared -f /etc/flared.conf --daemonize -p /var/run/flared.pid

Configure node setting

Connect index server by telnet as below.

$ telnet 10.0.0.1 12120

and check node states by stats command.

stats nodes

The result will be printed as below.

STAT 10.0.0.2:12121:role proxy
STAT 10.0.0.2:12121:state active
STAT 10.0.0.2:12121:partition -1
STAT 10.0.0.2:12121:balance 0
STAT 10.0.0.2:12121:thread_type 20
STAT 10.0.0.3:12121:role proxy
STAT 10.0.0.3:12121:state active
STAT 10.0.0.3:12121:partition -1
STAT 10.0.0.3:12121:balance 0
STAT 10.0.0.3:12121:thread_type 19
STAT 10.0.0.4:12121:role proxy
STAT 10.0.0.4:12121:state active
STAT 10.0.0.4:12121:partition -1
STAT 10.0.0.4:12121:balance 0
STAT 10.0.0.4:12121:thread_type 18
STAT 10.0.0.5:12121:role proxy
STAT 10.0.0.5:12121:state active
STAT 10.0.0.5:12121:partition -1
STAT 10.0.0.5:12121:balance 0
STAT 10.0.0.5:12121:thread_type 16
STAT 10.0.0.6:12121:role proxy
STAT 10.0.0.6:12121:state active
STAT 10.0.0.6:12121:partition -1
STAT 10.0.0.6:12121:balance 0
STAT 10.0.0.6:12121:thread_type 17
END

When you start flared for the first time, all node will treated as proxy.
So you should give role to each node by node role command.

node role 10.0.0.5 12121 master 1 0
node role 10.0.0.6 12121 slave 0 0
node role 10.0.0.3 12121 master 1 1
node role 10.0.0.4 12121 slave 0 1

and check node states.

stats nodes

The result:

STAT 10.0.0.2:12121:role proxy
STAT 10.0.0.2:12121:state active
STAT 10.0.0.2:12121:partition -1
STAT 10.0.0.2:12121:balance 0
STAT 10.0.0.2:12121:thread_type 20
STAT 10.0.0.3:12121:role master
STAT 10.0.0.3:12121:state active
STAT 10.0.0.3:12121:partition 1
STAT 10.0.0.3:12121:balance 1
STAT 10.0.0.3:12121:thread_type 19
STAT 10.0.0.4:12121:role slave
STAT 10.0.0.4:12121:state active
STAT 10.0.0.4:12121:partition 1
STAT 10.0.0.4:12121:balance 0
STAT 10.0.0.4:12121:thread_type 18
STAT 10.0.0.5:12121:role master
STAT 10.0.0.5:12121:state active
STAT 10.0.0.5:12121:partition 0
STAT 10.0.0.5:12121:balance 1
STAT 10.0.0.5:12121:thread_type 16
STAT 10.0.0.6:12121:role slave
STAT 10.0.0.6:12121:state active
STAT 10.0.0.6:12121:partition 0
STAT 10.0.0.6:12121:balance 0
STAT 10.0.0.6:12121:thread_type 17
END

Now, balance parametor of slave servers is 0.
If this parametor is 0, no request goes to the node.
But we want to execute read request on slave server.
So, we set balance parametor to 1, and devide read requests into halves for master and slave.

node role 10.0.0.6 12121 slave 1 0
node role 10.0.0.4 12121 slave 1 1

Now, all configuration has been finished.
Let's use flare.

$ telnet 10.0.0.2 12121

Set a key for test.

set key0 0 0 4
TEST

The result:

STORED

Get a key for test.

get key0

The result:

VALUE key0 0 4
TEST