Skip to content

Latest commit

 

History

History
164 lines (147 loc) · 6.06 KB

README.md

File metadata and controls

164 lines (147 loc) · 6.06 KB
Raw

Welcome to nopayloadclient!

Table of contents

Introduction

This client-side library is meant to communicate with NoPayloadDB (https://github.com/bnlnpps/nopayloaddb).

Setup

In the project folder, configure cmake, compile and install nopayloadclient

cmake -S . -B build -DCMAKE_INSTALL_PREFIX=<install_prefix> -DINCLUDE_DIR_NLOHMANN_JSON=<nlohmann_include_dir>
cmake --build build/
cmake --install build/

Omitting -DCMAKE_INSTALL_PREFIX=<install_prefix> will install it under the default directory (e.g. /usr/local/). Omitting -DINCLUDE_DIR_NLOHMANN_JSON=<nlohmann_include_dir> will use the header that ships with this product under contrib/.

If a custom installation prefix was defined, update the necessary environment variables:

export PATH=<install_prefix>/bin:$PATH
export LD_LIBRARY_PATH=<install_prefix>/lib:$LD_LIBRARY_PATH

Replace LD_LIBRARY_PATH by DYLD_LIBRARY_PATH on MacOS.

Configure the client by specifying the config file in an env variable

export NOPAYLOADCLIENT_CONF=/path/to/config.json

If this is not done, the parameters in config/default.json will be assumed.

Testing

The following command activates testing, reconfigures cmake, recompiles the code and runs the tests defined in the tests/ directory

cmake -DBUILD_TESTING=ON -S . -B build && cmake --build build/ && cd build/ && ctest --verbose && cd ..

Usage

To use this library in an independent project, include its header and the widely-spread nlohmann json header-only lib (https://github.com/nlohmann/json)

#include <nlohmann/json.hpp>
#include <nopayloadclient/nopayloadclient.hpp>

Firstly, create an instance of the client and configure it for a global tag

nopayloadclient::NoPayloadClient client;
client.setGlobalTag(<gt_name>);

Payloads can be inserted via

client.insertPayload(<pl_type>, <local_url>, <major_iov>, <minor_iov>);

and a dictionary mapping each payload type to one url can be retrieved via

client.getUrlDict(<major_iov>, <minor_iov>);

Every method returns a nlohmann::json object with a response code and a message body in the following format:

{'code': <code>, 'msg': <message>}

Currently, the following response codes are implemented:

0: ok
1: base error (no further details)
2: error when handling a payload
3: error in the database (server-side)

An example on how to use this tool in a c++ program can be found in examples/standalone.cpp. It can be compiled via

g++ -std=c++14 standalone.cpp -lnopayloadclient -lcurl -I../contrib/

Usage through command line interface

src/cli.cpp is an implementation of a command line interface. It is compiled along the rest of the project and can be run via

cli_npc <command> <parameters>

Some available commands are: get, createGlobalTag, createPayloadType, lockGlobalTag, unlockGlobalTag, deleteGlobalTag, insertPayload (overloaded), getSize, getPayloadTypes, getGlobalTags. Example workflow:

export NOPAYLOADCLIENT_CONF=/path/to/config.json
cli_npc checkConnection
cli_npc createGlobalTag example_gt
cli_npc getGlobalTags
cli_npc createPayloadType example_pt
cli_npc getPayloadTypes
cli_npc insertPayload example_gt example_pt /tmp/file.dat 7 0
cli_npc getUrlDict example_gt 11 0
cli_npc getSize

Configuration

Configuration of the client happens via a .json file. Examples can be found in the config/ dir. The default parameters are defined in config/default.json:

{
  "base_url": "localhost:8000",
  "api_res": "/api/cdb_rest/",
  "write_dir": "/tmp/remote_pl_store/",
  "read_dir_list": ["/tmp/remote_pl_store/", "/cvmfs/place/holder/"],
  "n_retries": 5,
  "retry_sleep_mean": 30,
  "cache_life_time": 10,
  "cache_max_mb": 1,
  "use_fake_backend": false,
  "logger": "terminal",  // choose from ["terminal", "syslog"]
  "log_level": "INFO"
}

base_url and api_res together define the url to the rest api of the back end. The example above assumes that the service can be reached under localhost:8000/api/cdb_rest/.

write_dir is where the payload files are copied to. To optimise the performance of the file system, subdirectories are created based on the checksum of files. The naming is not meant to be human-readable.

read_dir_list defines a list of directories that the client will use to find a payload file. The directories are checked in the defined order and the complete path of the first payload file found will be used as the payload url. In the simplest case, write_dir would be the only entry in read_dir_list.

n_retries specifies the number of retries to contact the rest api before throwing an error. The waiting time before each retry grows exponentially.

retry_sleep_mean determines the mean sleep time after a request failed with a connection time out. The actual sleep time follows a uniform random distribution.

cache_life_time is the lifetime of each entry in the cache. Set to -1 for infinite lifetime.

cache_max_mb is the maximum (approximated) size in mb that the cache will occupy. The oldest entry will be removed until the cache is below the limit again. Set to -1 for unlimited size.

use_fake_backend: if set to true, the client will not contact the actual rest api but instead simulate its response assuming a single, minimal global tag. This is useful when testing the client's interface (e.g. when integrating into an existing software framework) in an environment where the backend is not reachable. Allows only read operations. Some details on the simulated DB content: One global tag ExampleGT, two payload lists ExamplePT1 and ExamplePT2 with one IOV each that is valid after major_iov=42 and minor_iov=42. The inserted payload files where called example_file.dat and contained the string data.

logger: choose where the logfiles should be written. options: [terminal, syslog]. Default is terminal.

log_level: standard convention of log levels. options: [DEBUG, INFO, WARNING, ERROR]. Default is WARNING.