The YaCy Grid is the second-generation implementation of YaCy, a peer-to-peer search engine. A YaCy Grid installation consists of a set of micro-services which communicate with each other using a common infrastructure for data persistence. The required storage functions of the YaCy Grid are:
All YaCy components are microservices which can be deployed i.e. using Docker or other application hosting methods which are provided as cloud services from many different providers. A YaCy grid will scale with the number of microservices that connect to a common broker. That broker is the MCP, this application.
The MCP provides http servlets to access the services mentioned above: assets, messages, databases. If a YaCy Grid microservice wants to connect to this infrastructure, it connects to the MCP to use them. The API to access MCP servlets are integrated into the MCP as well, every YaCy Grid service must integrate the whole MCP as infrastructure to get access to the API. This makes it possible that the MCP acts also as router to the infrastructure: if a YaCy Grid service has used the MCP once, it learns from the MCP to connect to the infrastucture itself. For example:
The asset storage, message distribution and database services provided by the MCP are handled in an abstract way, that means there is no need to know which system actually performs the operation. However, the following external services are supposed to be supported by the MCP:
For a full enterprise-integration installation, the operator has to start and support a ftp server, a rabbitmq and an elasticsearch. However, all of them can be omitted and the MCP will provide all of these services with built-in functions. For database operations and message storage a MapDB is used, asset files are stored in a local data directory.
Every YaCy Grid service has a default port. However every service can be started several times. The service is able to detect that another service is running on the same port and will choose an alternative port. That means we will not have any port number collisions any more, just more microservices for one grid.
The default port number of the MCP is 8100
Other port numbers will be:
Please join our forums at https://searchlab.eu
At this time, yacy_grid_mcp is not provided in compiled form, you easily build it yourself. It's not difficult and done in one minute! The source code is hosted at https://github.com/yacy/yacy_grid_mcp, you can download it and run loklak with:
> git clone https://github.com/yacy/yacy_grid_mcp.git > cd yacy_grid_mcp > gradle run
Maven and Gradle tools should be installed. To refresh gradle settings in eclipse, right click on project -> configure -> Add Gradle Nature
Download apache-ftpserver-1.1.1.tar.gz (or a later version) from https://mina.apache.org/downloads-ftpserver.html
decompress the server package with
tar xfz apache-ftpserver-1.1.1.tar.gz
modify the write right of the anonymous user that we will use to write assets: edit the file
set the write permission for anonymous.. if you don't want to set an user account
to create a user account, clone all entries for the user anonymous, rename the account name and set a password using a md5 encoding. The following command will encode the password 'password4account':
echo -n password4account | md5sum
set a home path for the user, i.e.
set a high number of allowed connections for
The complete configuration may then look like
ftpserver.user.yacygrid.userpassword=<here is the md5sum> ftpserver.user.yacygrid.homedirectory=./res/home ftpserver.user.yacygrid.enableflag=true ftpserver.user.yacygrid.writepermission=true ftpserver.user.yacygrid.maxloginnumber=2000 ftpserver.user.yacygrid.maxloginperip=2000 ftpserver.user.yacygrid.idletime=30000 ftpserver.user.yacygrid.uploadrate=0 ftpserver.user.yacygrid.downloadrate=0
edit the file apache-ftpserver-1.1.1/bin/ftpd.sh and set JAVA_HOME according to your system, i.e. on a Mac you set
run the server, doing:
cd apache-ftpserver-1.1.1 bin/ftpd.sh res/conf/ftpd-typical.xml
This will run the ftp server at port 2121. To test the connection use a standard ftp client and start it with
> ftp -P 2121 [email protected]
Download rabbitmq-server-generic-unix-3.6.6.tar.xz (from https://www.rabbitmq.com/download.html)
Extract the package with
tar xf rabbitmq-server-generic-unix-3.6.6.tar.xz
install the erlang programming language, rabbitmq is written in erlang. I.e. in debian, install:
apt-get install erlang
if you own a Mac, install brew and then install erlang with
brew install erlang
Run the server with
install the management plugin to be able to use a web interface:
rabbitmq_server-3.6.6/sbin/rabbitmq-plugins enable rabbitmq_management
add a rabbitmq user 'anonymous' with password 'yacy':
rabbitmq_server-3.6.6/sbin/rabbitmqctl add_user anonymous yacy
set administration rights to that user (without that, it is not possible to open the administration pages)
rabbitmq_server-3.6.6/sbin/rabbitmqctl set_user_tags anonymous administrator
set access right to vserver path '/' (makes it possible that queues are written over the api)
rabbitmq_server-3.6.6/sbin/rabbitmqctl set_permissions -p / anonymous "." "." ".*"
to make it possible that the rabbitmq server can be accessed from outside of localhost, add a configuration file to
rabbitmq_server-3.6.6/etc/rabbitmq/rabbitmq.config with the following content:
if you made it possible to set a remote connection to localhost, change the default password of the guest account (or remove the guest account)
To view the administration pages, open in your web browser: http://127.0.0.1:15672
open http://127.0.0.1:15672/api/ for an api documentation
open http://127.0.0.1:15672/cli/ for the rabbitmqadmin
The mcp will create a subdirectory
data/mcp-8100. There, within a
conf sub-path you can place a file
config.properties which has the same structure as the file in
<application-home>/conf/config.properties. Just copy that file to
data/mcp-8100/conf/config.properties and replace the default values with your own.
You should set the ftp address and the broker address using url-encoded user-name/pw-settings, like:
grid.ftp.address = <user>:<pw>@<ftp-host>:2121 grid.broker.address = <user>:<pw>@<broker-host>:5672
To test the api, try the following example:
This will send a message "hello_world" to the queue 'test' of service 'test'. You can ask for the number of entries in the queue with
To get an entry from such a queue, call:
If you did not run the RabbitMQ message server, the messages are written to
If you started a RabbitMQ message server, please monitor the writing of the messages at the web page http://127.0.0.1:15672
This will write an asset to the path xx/test.txt with the content "hello_world".
will load the asset again.
If you started a ftp server, the file(s) will be written relatively to the root path of the ftp home path. If you did not start a ftp server, you can find the file in data/mcp-8100/assets/xx/test.txt
The MCP organises the connection to the remote RabbitMQ server and a ftp server, but if another Grid Service is started, the MCP tells that service to handle the connection to the RabbitMQ and ftp server itself. The MCP can act as such an external service: just start another MCP and it will run at port 8101. Repeat the curl commands as given in the example above, but now use port 8101 to access the MCP. You will see that the second MCP learns from the first MCP to handle the connection by itself.
To install yacy_grid_mcp with Docker please refer to the yacy Docker installation readme.
To install yacy_grid_mcp on Cloud Providers please look documentations at yacy_grid_mcp Cloud installation readme.
wget https://www.land.nrw/ --warc-file=land.nrw
gzip -9 land.nrw
curl -X POST -F "[email protected];type=application/octet-stream" -F "flatfile=true" -o land.nrw.flatjson http://127.0.0.1:8500/yacy/grid/parser/parser.json
This is a community project and your contribution is welcome!
LGPL 2.1 (C) by Michael Peter Christen
Have fun! @0rb1t3r