Tempo: Like Prometheus, but for logs.

Design doc

Tempo is a horizontally-scalable, highly-available, multi-tenant, log aggregation system inspired by Prometheus. It is design to be very cost effective, as it does not index the contents of the logs, but rather a set of labels for each log steam.

Usage Instructions

Tempo is running in the ops-tools1 cluster. You can query logs from that cluster using the following commands:

$ go get github.com/grafana/tempo/cmd/logcli
$ . $GOPATH/src/github.com/grafana/tempo/env # env vars inc. URL, username etc
$ logcli labels job
https://logs-dev-ops-tools1.grafana.net/api/prom/label/job/values
cortex-ops/consul
cortex-ops/cortex-gw
...
$ logcli query '{job="cortex-ops/consul"}'
https://logs-dev-ops-tools1.grafana.net/api/prom/query?query=%7Bjob%3D%22cortex-ops%2Fconsul%22%7D&limit=30&start=1529928228&end=1529931828&direction=backward&regexp=
Common labels: {job="cortex-ops/consul", namespace="cortex-ops"}
2018-06-25T12:52:09Z {instance="consul-8576459955-pl75w"} 2018/06/25 12:52:09 [INFO] raft: Snapshot to 475409 complete
2018-06-25T12:52:09Z {instance="consul-8576459955-pl75w"} 2018/06/25 12:52:09 [INFO] raft: Compacting logs from 456973 to 465169

The logcli command is temporary until we have Grafana integration. The URLs of the requests are printed to help with integration work.

$ logcli help
usage: logcli [<flags>] <command> [<args> ...]

A command-line for tempo.

Flags:
  --help         Show context-sensitive help (also try --help-long and --help-man).
  --addr="https://log-us.grafana.net"
                 Server address.
  --username=""  Username for HTTP basic auth.
  --password=""  Password for HTTP basic auth.

Commands:
  help [<command>...]
    Show help.

  query [<flags>] <query> [<regex>]
    Run a LogQL query.

  labels <label>
    Find values for a given label.

$ logcli help query
usage: logcli query [<flags>] <query> [<regex>]

Run a LogQL query.

Flags:
  --help         Show context-sensitive help (also try --help-long and --help-man).
  --addr="https://log-us.grafana.net"
                 Server address.
  --username=""  Username for HTTP basic auth.
  --password=""  Password for HTTP basic auth.
  --limit=30     Limit on number of entries to print.
  --since=1h     Lookback window.
  --forward      Scan forwards through logs.

Args:
  <query>    eg '{foo="bar",baz="blip"}'
  [<regex>]

API

nb Authentication is out of scope for this project. You are expected to run an authenticating reverse proxy in front of our services, such as an Nginx with basic auth or a OAuth2 proxy.

There are 4 API endpoints:

• POST /api/prom/push

  For sending log entries, expects a snappy compresses proto in the HTTP Body.

• GET /api/prom/query

  For doing queries, accepts the following paramters in the query-string:

  • query: a logQL query
  • limit: max number of entries to return
  • start: the start time for the query, as a nanosecond Unix epoch (nanoseconds since 1970)
  • end: the end time for the query, as a nanosecond Unix epoch (nanoseconds since 1970)
  • direction: forward or backward, useful when specifying a limit
  • regexp: a regex to filter the returned results, will eventually be rolled into the query language

  Responses looks like this:

  {
    "streams": [
      {
        "labels": "{instance=\"...\", job=\"...\", namespace=\"...\"}",
        "entries": [
          {
            "timestamp": "2018-06-27T05:20:28.699492635Z",
            "line": "..."
          },
          ...
        ]
      },
      ...
    ]
  }
  

• GET /api/prom/label

  For retrieving the names of the labels one can query on.

  Responses looks like this:

  {
    "values": [
      "instance",
      "job",
      ...
    ]
  }
  

• GET /api/prom/label/<name>/values For retrieving the label values one can query on.

  Responses looks like this:

  {
    "values": [
      "default",
      "cortex-ops",
      ...
    ]
  }