Gaurun is a general push notification server written in Golang. It proxies push requests to APNs and FCM and asynchronously executes them via HTTP/2. It helps you when you need to bulkly sends push notification to your users (e.g., when you need to exec 10 million push at once!) or when some other API server which must response quickly needs to push. Since it leverages Golang's powerful concurrent feature, it gives high performance.
In addition to performance, it's important not to lost pushes over sever crashes or hardware failures. Gaurun can use its access log for kind of transaction journal and can re-push only failed notification later (We provide a special command for this. See Usage).
Currently we support the following platforms:
Production ready.
There are two way to install Gaurun; using a precompiled binary or install from source. Downloading a precompiled binary is easiest and recommended.
To install a precompiled binary, download the appropriate zip package for your OS and architecture from here. Once the zip is downloaded, unzip it and place the binary where you want to use (if you want to access it from the command-line, make sure to put it on $PATH).
To compile from source, you need Go1.13 or later. After setup, then clone the source code by running the following command,
$ git clone https://github.com/mercari/gaurun.gitTo fetch dependencies and build, run the following make tasks,
makeTo run gaurun, you must provide configuration path via -c option (See CONFIGURATION.md about details),
$ bin/gaurun -c conf/gaurun.tomlUse -help to see more options.
Gaurun can recover from server crashes or hardware failures while pushing. It can use its access log for kind of transaction journal and can re-push only failed notifications later. We provide the special command for this, use it like the following (assuming that access log is generated to /tmp/gaurun.log),
$ bin/gaurun_recover -c conf/gaurun.toml -l /tmp/gaurun.logSee CONFIGURATION.md about details.
NOTE: The default configuration is just for development and not high performant. For production usage, tune the performance with some parameters such as workers and queues and pusher_max in the core section.
API specification is defined on SPEC.md.
- Tatsuhiko Kubo(@cubicdaiya)
- Masahiro Sano(@kazegusuri)
- Taichi Nakashima(@tcnksm)
Please read the CLA below carefully before submitting your contribution.
Copyright 2014-2017 Mercari, Inc.
Licensed under the MIT License.