1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
|
## cloudant-dbcore
### Overview
Cloudant-dbcore is a highly available, fault-tolerant, clustered, fully api-compliant version of [Apache CouchDB][1]. While it appears to the end-user as one CouchDB instance, it is in fact one or more nodes in an elastic cluster, acting in concert to store and retrieve documents, index and serve views, and serve CouchApps. Dbcore has been developed and is continually maintained by [Cloudant][2] who offer hosted CouchDB as a service.
Clusters behave according to concepts outlined in [Amazon's Dynamo paper][4], namely that each node can accept requests, data is placed on partitions based on a consistent hashing algorithm, and quorum protocols are for read/write operations.
### Contents
* README.md this file
* LICENSE open-source license governing dbcore
### Getting Started
#### Prerequisites
Cloudant-dbcore has the same dependencies as CouchDB:
* Erlang (R13B03 or higher)
* ICU (4.2 is preferable)
* Spidermonkey (1.9.2 preferable, [https://launchpad.net/~commonjs/+archive/ppa/][6])
* LibCurl
* OpenSSL
* make
* Python (2.4 or higher)
#### Installing prerequisites on Ubuntu
sudo apt-get install erlang libicu42 libcurl-openssl-dev
To install Spidermonkey 1.9.2 from PPA:
sudo apt-key adv --keyserver keyserver.ubuntu.com --recv-keys 74EE6429
sudo bash -c 'echo "deb http://ppa.launchpad.net/commonjs/ppa/ubuntu karmic main" >> /etc/apt/sources.list'
sudo apt-get update
sudo apt-get install libmozjs-1.9.2
sudo ln -s /usr/lib/libmozjs-1.9.2.so /usr/lib/libmozjs.so
#### Installing prerequisites on Mac OS X with [Homebrew][7]
brew install erlang icu4c spidermonkey
brew ln icu4c
#### Building and installing dbcore
`$CLOUDANT_SRC` is the directory holding your downloaded source files, while `$PREFIX` is the prefix to which the software is installed (defaults to `/opt/dbcore`):
cd $CLOUDANT_SRC
./configure -p $PREFIX
make
sudo make install
`sudo` is only necessary when installing to a prefix which is not user-writeable. In any case, the installer tries to chown the database directory and logfile to the user who configured dbcore.
#### Starting dbcore
$PREFIX/bin/dbcore
Now, visit http://localhost:5984/_utils in a browser to verify the CouchDB node is operational.
dbcore listens on two ports. Defaults and explanations:
* 5984 - front door, cluster-aware port, appears as a standalone CouchDB.
* 5986 - back door, single-node port, used for admin functions
Note: see the rel/sv/README file for information on using runit to stop/start dbcore.
#### Joining a new node to the cluster
Each dbcore node has a local `nodes` database, accessible through the backend interface on port 5986. Documents in the `nodes` DB name nodes in the cluster. To add a new node, create a document with that node's name as the ID. For example
curl -X PUT http://foo.example.com:5986/nodes/dbcore@bar.example.com -d {}
Everything else should be automatic, provided the machines can ping each other and the nodes set the same magic cookie. You are advised to change the magic cookie from the default in `rel/etc/vm.args` when on a public network.
#### Local development cluster
The `make dev` target will build a three-node cluster under the rel/ directory. Get the nodes running, like above, by doing the following (in separate terminals):
./rel/dev1/bin/dbcore
./rel/dev2/bin/dbcore
./rel/dev3/bin/dbcore
These development nodes listen on ports 15984/15986 (dev1), 25984/25986 (dev2), and 35984/35986 (dev3). Now, once the nodes are started, join the dev2 node by sending this PUT to dev1's listening backend port:
curl -X PUT http://127.0.0.1:15986/nodes/dev2@127.0.0.1 -d {}
To verify the two-node cluster has been linked properly, on either node (via proper frontend port), try:
curl http://127.0.0.1:15984/_membership
You should see something similar to this:
{"all_nodes":["dev1@127.0.0.1","dev2@127.0.0.1"],"cluster_nodes":["dev1@127.0.0.1","dev2@127.0.0.1"]}
Add node 3 to the cluster by sending a similar PUT to either of the first two nodes.
#### Now What?
If the above steps were successful, you should have a running dbcore cluster that looks just like a standalone CouchDB. You may interact with it the same way you would a standalone CouchDB, via the HTTP REST interface.
Because every node can handle requests equally, you may want to put a load balancer in front of the cluster and set up a round-robin strategy for distributing incoming requests across all of your cluster's nodes.
##### Create a database:
curl -X PUT http://loadbalancer:5984/test_db
Also note that 'q' and 'n' are query string arguments that may be specified. These are Cloudant-specific options, and their values and defaults are discussed in the Configuration section below. Ex: &q=12 or &n=4.
##### Create a document:
curl -X PUT http://loadbalancer:5984/test_db/doc_1 -d '{"a":1,"b":2}'
You may also provide 'r' and 'w' on the GET and PUT/POST document operations respectively. Their values and defaults are discussed in the Configuration section below. Ex: &r=3 for high consistency reads, or &w=1 for higher throughput writes.
##### Check out Futon, CouchDB's web UI:
* [http://loadbalancer:5984/_utils][5]
### Configuration
#### Cluster constants
_Q_ - number of partitions per database. Q is specified in the default.ini file, but may be provided as a URL parameter when creating a database. Default value is 8, query parameter is &q=12 or similar.
_N_ - replication constant. N defaults to 3, but can vary by database just as Q can vary. N copies of each document will be written to the data store, on N different nodes.
_R_ - read quorum constant. N writes have occurred for each document, as noted above. When reads are requested, N reads are sent to the N nodes that store the particular document. The system will return to the requesting client with the document when R successful reads have returned, and agree on versioning. R defaults to 2. Lower R values often result in faster reads at the expense of consistency. Higher R values usually result in slower reads, but more consistent, or agreed-upon data values returning.
_W_ - write quorum constant. When writing the N copies, the data store will respond to the write client after W successful writes have completed. The remaining N-W writes are still being attempted in the background, but the client receives a 201 Created status and can resume execution. W defaults to 2. Lower W values mean more write throughput, and higher W values mean more data durability.
### Troubleshooting
#### `make install` fails with "read file info /usr/lib/erlang/man/man5/modprobe.d.5 failed"
You can fix this by removing that file:
`sudo rm /usr/lib/erlang/man/man5/modprobe.d.5`
#### Dbcore fails on startup with a "Can't set long node name!" error
Cloudant-dbcore requires that your system be set up with a fully qualified domain name. Make sure that the command `hostname -f` outputs a full hostname. If not, add the full hostname of your server to the `/etc/hosts` file.
### Contact
Cloudant folks are usually hanging out in IRC. Freenode, channel #cloudant. We may also be reached:
* [http://cloudant.com][2]
* [info@cloudant.com][3]
----
[1]: http://couchdb.apache.org
[2]: http://cloudant.com
[3]: mailto:info@cloudant.com
[4]: http://www.allthingsdistributed.com/2007/10/amazons_dynamo.html
[5]: http://loadbalancer:5984/_utils
[6]: https://launchpad.net/~commonjs/+archive/ppa/
[7]: http://mxcl.github.com/homebrew/
|
