https://github.com/aqtrans/gorram

A simple gRPC-powered monitoring system written in Go
https://github.com/aqtrans/gorram

grpc monitoring monitoring-server nagios sensu

Last synced: 2 months ago
JSON representation

A simple gRPC-powered monitoring system written in Go

• Host: GitHub
• URL: https://github.com/aqtrans/gorram
• Owner: aqtrans
• License: mit
• Created: 2018-07-22T03:51:14.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2025-04-20T10:43:47.000Z (12 months ago)
• Last Synced: 2025-04-20T12:09:41.352Z (12 months ago)
• Topics: grpc, monitoring, monitoring-server, nagios, sensu
• Language: Go
• Size: 562 KB
• Stars: 1
• Watchers: 3
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Gorram 

*So you know when the primary buffer panel falls off your gorram ship*

Written in Go. This is my attempt at a creating a monitoring system, similar to Sensu or Nagios. 

Inspired by Telegraf, a single-binary metrics agent, the goal is to have no external dependencies or process forking unless absolutely necessary. 

A vast majority of the current checks are implemented using the same library as Telegraf, [gopsutil](github.com/shirou/gopsutil). 

The server holds all configuration information in a YAML file on a per-client basis, and that client config is pushed to the client upon connect. 

[fsnotify](https://github.com/fsnotify/fsnotify) is used to watch and reload and send the config to the client as necessary, as part of the ping/heartbeat. 

Formerly using gRPC, I've recently moved to using [Twirp](https://github.com/twitchtv/twirp). Replacing Mutual TLS authentication with ed25519 signature verification. Each client has a private key specified in their `client.yml`, with their public key in the server's `conf.d/$clientname.yml`. On client connection, the shared secret is signed and sent to the server, to be verified against that client's public key. 

## Quickstart:  

- Fetch the package itself: `go get git.jba.io/go/gorram`  

- Or clone from git: `git clone https://git.jba.io/go/gorram`  

- Build: `./build.sh build`  

- Move/edit `client.yml.dist`, `clientname.yml.dist`, and `server.yml.dist` as necessary: 

    - `server.yml` is stored on the server-side

    - `client.yml` is stored on the client-side 

    - `$client_name.yml` files are stored in a `conf.d` directory on the server-side

        - One yml file per-client, with the filename matching the literal `$client_name` 

    - `secret_key` in `client.yml` and `server.yml` must match, it is the shared secret signed and verified

- Generate public/private keys for the client: `./gorram-client -generate-keys`

    - Place the private key into `client.yml` as `private_key` on the client-side

    - Place the public key into `conf.d/$client_name.yml` as `public_key` on the server-side

After 60 seconds, the checks should start flowing, checking anything specified in `$client_name.yml` on the client. 

## Config files:

- `client.yml.dist` is an example of the client-side configuration file

- `clientname.yml.dist` is an example of a server-side, client-specific configuration file, storing all the checks for the client

- `server.yml.dist` is an example of the server-side configuration file

## Currently implemented checks:

- Deluge: max number of torrents in an error, checking, or downloading state.  

- Disk Space: max percentage of disk space used on multiple mounts.  

- Load Average: max load average.  

- Process Existence: check that a given process is running, full path to the binary.  

- HTTP GET: checks that a specified URL returns a 200, and optionally, checks that the response body matches a given string.  

- Memory Usage: check percentage of used memory

- Postgres Replication: query the `pg_stat_replication` table on a Postgres master for a specific `client_addr`. 

Postgres check needs some aditional setup using `psql`:  

- To create user: `create user gorram with password '[reallySecurePassword]';`  

- To grant user permissions to monitor replication status: `grant pg_monitor to gorram;`  

## Alerts:  

Alerts will be sent if the number of identical alert occurences is 1, greater than 5, or the occurrences is evenly divisible by 10.  

This is a very rudimentary 'backoff' implementation. 

## Todo:  

- [x] Add additional alerting mechanisms (Pushover, email, etc). 

    - Currently [Pushover](https://pushover.net/) push notifications and log are implemented. 

- [ ] Add client expiration/deletion.  

    - Right now if a client disappears, the server will start alerting and never stop until the client re-appears. 

- [ ] Related to the above, implement some kind of frontend to interact with the server; manually delete clients, reload config, etc.  

    - CLI is in the works, but still working on the gRPC server endpoints

    - Basic web UI has been implemented, currently used to list and mute alerts