cloud-crawler

cloud-crawler is a simple web crawler designed to be deployed on cloud servers.

see project group @ https://team.oschina.net/cloudwalk

Basic structure

In current design, the cloud-crawler has one master-node and multiple slave-nodes which would all be deployed on cloud servers(such as AWS and AliYun).

The graph below shows a sketch of the basic project structure, note that the structure may be modified if we find it not applicable or a better design comes out.

• The master-node contains the coordinator module and databases. The coordinator accesses the crawling queue in Key-Value database and dispatches work tasks to the crawlers on slave-nodes, this module is designed to handle multiple slave-nodes' requests(which means the slave-node should be easily added or removed to ensure flexible extension).
• There are two kinds of databases installed on master-node, in current design they are Redis and MySQL. While Redis-DB serves as the working queue and replaces the Berkeley-DB in the original crawler4j project, MySQL-DB stores relatively fixed data such as slave-nodes' status, crawling task status, etc.
• The crawler on slave-nodes uses a few modules from the crawler4j project on github, with its controller and database access changed in order to work as part of the distributed system. A communication module is implemented so that the crawlers can link up with the coordinator on the master-node.
• There are FTP servers on slave-nodes so that the user may get the crawling products(the pages crawled will be processed on local slave-node, and if there are any processing results, they will be pushed to the FTP server waiting for download).

Brief deployment guide(On Linux server)

Normal way

See complete version here.

• Set up FTP service so that file can be transferred.
• Install java runtime environment and add environment variables.
• (Ignore this step on slave-node) Install and configure Redis and MySQL services.
• Upload the project's jar file(The program for master-node and slave-node will be packaged in one jar file) and "cloud-crawler.sh"(under "deployment/shell/") to servers, execute "cloud-crawler.sh" with different arguments in Linux shell to start each service.
  $ mv -iuv /usr/rootftp/cloud-crawler-jar-with-dependencies.jar /usr/crawler
$ chmod +x cloud-crawler.sh
$ ./cloud-crawler.sh -m 或 cloud-crawler.sh -s 或 ./cloud-crawler.sh -s IP PORT CODE

Using docker(NOT AVAILABLE YET)

• Set up master node as mentioned above.
• Get slave's docker image at release page or under "deployment/app/".
• Deploy slave-nodes on using the docker image on as many servers as you want.

Optional objectives

These following features might be implemented if possible:

• Create a docker image containing the environment of slave-node for fast deployment and extension.
• ~~Use web pages to show the current status and progress of the crawlers.~~√
• ...(more to be added)

Development guide

API guide (Request from Front-end)

Request: /slave.json

Response:

{
  "hostname": "localhost",
  "ip": "127.0.0.1",
  "port": -1,
  "name": "slave0",
  "number": 0
}

Request: /slaves.json

Response:

{
  "slave0": {
    "hostname": "localhost",
    "ip": "127.0.0.1",
    "port": -1,
    "name": "slave0",
    "number": 0
  },
  "slave1": {
    "hostname": "localhost",
    "ip": "127.0.0.1",
    "port": -1,
    "name": "slave1",
    "number": 0
  },
  "slave2": {
    "hostname": "remote",
    "ip": "127.0.0.1",
    "port": -1,
    "name": "slave0",
    "number": 0
  },
  "slave3": {
    "hostname": "remote",
    "ip": "127.0.0.1",
    "port": -1,
    "name": "slave1",
    "number": 0
   }
}

Request: /number.json

Response:

{
  "slave0": 1,
  "slave1": 2,
  "slave2": 3,
  "slave3": 4
}