Pulse is a stat collecting and publishing service. It serves historical stats over an http api while live stats are sent to mist for live updates.
Simply running pulse -s
will start pulse with the default config options.
pulse -h
or pulse --help
will show more detailed usage and config options:
Usage:
pulse [flags]
Flags:
-a, --aggregate-interval int Interval at which stats are aggregated (default 15)
-b, --beat-interval int Heartbeat frequency (seconds) (default 30)
-c, --config-file string Config file location for server
-C, --cors-allow string Sets the 'Access-Control-Allow-Origin' header (default "*")
-H, --http-listen-address string Http listen address (default "127.0.0.1:8080")
-i, --influx-address string InfluxDB server address (default "http://127.0.0.1:8086")
-I, --insecure Run insecure (default true)
-k, --kapacitor-address string Kapacitor server address (http://127.0.0.1:9092)
-l, --log-level string Level at which to log (default "INFO")
-m, --mist-address string Mist server address
-M, --mist-token string Mist server token
-p, --poll-interval int Interval to request stats from clients (default 60)
-r, --retention int Number of weeks to store aggregated stats (default 1)
-s, --server Run as server
-S, --server-listen-address string Server listen address (default "127.0.0.1:3000")
-t, --token string Security token (recommend placing in config file) (default "secret")
-v, --version Print version info and exit
{
"server": true,
"server-listen-address": "127.0.0.1:3000",
"http-listen-address": "127.0.0.1:8080",
"influx-address": "http://127.0.0.1:8086",
"kapacitor-address": "http://127.0.0.1:9092",
"insecure": true,
"mist-address": "",
"mist-token": "",
"log-level": "info",
"cors-allow": "*",
"token": "secret",
"poll-interval": 60,
"aggregate-interval": 15,
"beat-interval": 30,
"retention": 12
}
Route | Description | Output |
---|---|---|
GET /keys | Returns list of stats being recorded | string array |
GET /tags | Returns list of filterable tags | string array |
GET /latest/{stat}* | Returns latest stat (averages if multiple filters applied) | json stat object |
GET /hourly/{stat}** | Returns hourly averages for stat | json array of stat objects |
GET /daily/{stat}** | Returns average for stat at the same daily time | string map |
ALERTS (requires "kapacitor-address" to be configured)
Route | Description | Payload | Output |
---|---|---|---|
POST /alerts | Add a kapacitor alert | json alert object | json alert object |
PUT /alerts | Update a kapacitor alert | json alert object | json alert object |
DELETE /alerts/{alert} | Delete a kapacitor alert | nil | success message |
*
: reserved query parameters is 'verb', all others act as filters
**
: reserved query parameters are 'backfill', 'verb', 'start', and 'stop', all others act as filters
note: The API requires a token to be passed for authentication by default and is configurable at server start (--token
). The token is passed in as a custom header: X-AUTH-TOKEN
.
For examples, see the api's readme.
json:
{
"time": 1465419600,
"value": 0.75
}
Fields:
- time: Unix epoch timestamp of stat
- value: Numeric value of stat
json:
{
"tags": {"host":"abc"},
"metric": "cpu_used",
"level": "crit",
"threshold": 80,
"duration": "30s",
"post": "http://127.0.0.1/alert"
}
Fields:
- tags: Populates the WHERE
- metric: Stat to track
- level: Alert level (info, warn, crit)
- threshold: Limit that alert is triggered at
- duration: How far back to average (5m)
- post: Api to hit when alert is triggered
A pulse relay is a service that connects to pulse and advertises stats that are available for collection. A relay implementation is available in the pulse project and can be embedded in other projects.
For an example, look in the README for relay.
The TCP api used to communicate between the pulse server and a relay is simple and is designed to be human readable and debuggable. It is newline delimited.
Command | Description | Response |
---|---|---|
id {id} |
Must be the first command to be run, identifies the client to the server | ok |
add {name} |
Exposes a stat that can be collected by the server | ok |
remove {name} |
Removes a stat previously exposed to the server | ok |
Command | Description | Response |
---|---|---|
get {tag,tag2} |
Request a list of stats corrosponding to the list of tags passed in | got {tag:value} |
flush |
Clear all current values from the stat collectors | ok |
override {duration} {tag:interval} |
for duration seconds, bump the collection interval from the default to interval for each tag:interval |
ok |
- If an override is specified for a stat, and a new machine comes online and connects, that override is NOT honored.
- Pulse server does not actively connect to servers to have stats pushed to it, rather, it waits for stat collecting machines to connect and then requests certain stats on specific intervals.
Contributions to the pulse project are welcome and encouraged. Pulse is a Nanobox project and contributions should follow the Nanobox Contribution Process & Guidelines.
Mozilla Public License Version 2.0