The Timber library works directly with the standard Elixir Logger and installs itself as a backend during the setup process. In this way, basic logging is no different than logging without Timber:

Logger.debug("My log statement")
Logger.info("My log statement")
Logger.warn("My log statement")
Logger.error("My log statement")

• Search it with queries like: error message
• Alert on it with threshold based alerts
• View this event's metadata and context

...read more in our docs

Log structured data without sacrificing readability:

event_data = %{customer_id: "xiaus1934", amount: 1900, currency: "USD"}
Logger.info("Payment rejected", event: %{payment_rejected: event_data})

• Search it with queries like: type:payment_rejected or payment_rejected.amount:>100
• Alert on it with threshold based alerts
• View this event's data and context

...read more in our docs

Add shared structured data across your logs:

Timber.add_context(build: %{version: "1.0.0"})
Logger.info("My log message")

• Search it with queries like: job.id:123
• View this context when viewing a log's metadata

...read more in our docs

Time code blocks:

timer = Timber.start_timer()
# ... code to time ...
Logger.info("Processed background job", event: %{background_job: %{time_ms: timer}})

Log generic metrics:

Logger.info("Processed background job", event: %{background_job: %{time_ms: 45.6}})

• Search it with queries like: background_job.time_ms:>500
• Alert on it with threshold based alerts
• View this log's metadata in the console

...read more in our docs

Note: This tip refers to traditional background jobs backed by a queue. For native Elixir processes we capture the context.runtime.vm_pid automatically. Calls like spawn/1 and Task.async/1 will automatially have their pid included in the context.

For traditional background jobs backed by a queue you'll want to capture relevant job context. This allows you to segement logs by specific jobs, making it easy to debug and monitor your job executions. The most important attribute to capture is the id:

%Timber.Contexts.JobContext{queue_name: "my_queue", id: "abcd1234", attempt: 1}
|> Timber.add_context()

Logger.info("Background job execution started")
# ...
Logger.info("Background job execution completed")

• Search it with queries like: background_job.time_ms:>500
• Alert on it with threshold based alerts
• View this log's metadata in the console

...read more in our docs

We use this trick internally at Timber to track communication with external services. It logs requests and responses to external services, giving us insight into response times and failed requests.

Below is a contrived example of submitting an invoice to Stripe.

alias Timber.Events.HTTPRequestEvent
alias Timber.Events.HTTPResponseEvent

method = :get
url = "https://api.stripe.com/v1/invoices"
body = "{\"customer\": \"cus_BHhZyYRirFrPkz\"}"
headers = %{}

Logger.info fn ->
  event = HTTPRequestEvent.new(direction: "outgoing", service_name: "stripe", method: method, url: url, headers: headers, body: body)
  message = HTTPRequestEvent.message(event)
  {message, [event: event]}
end

timer = Timber.start_timer()
case :hackney.request(method, url, headers, body, with_body: true) do
  {:ok, status, resp_headers, resp_body} ->
    Logger.info fn ->
      event = HTTPResponseEvent.new(direction: "incoming", service_name: "stripe", status: status, headers: resp_headers, body: resp_body, time_ms: Timber.duration_ms(timer))
      message = HTTPResponseEvent.message(event)
      {message, [event: event]}
    end

  {:error, error} ->
    message = Exception.message(error)
    Logger.error(message, event: error)
    {:error, error}
end

Note: Only method is required for HTTPRequestEvent, and status for HTTPResponseEvent. body, if logged, will be truncated to 2048 bytes for efficiency reasons. This can be adjusted with Timber.Config.http_body_size_limit/0.

• Search it with queries like: background_job.time_ms:>500
• Alert on it with threshold based alerts
• View this log's metadata in the console

...read more in our docs

By default, Timber will capture and structure all of your errors and exceptions, there is nothing additional you need to do. You'll get the exception message, name, and backtrace. But, in many cases you need additional context and data. Timber supports additional fields in your exceptions, simply add fields as you would any other struct.

defmodule StripeCommunicationError do
  defexception [:message, :customer_id, :card_token, :stripe_response]
end

raise(
  StripeCommunicationError,
  message: "Bad response #{response} from Stripe!",
  customer_id: "xiaus1934",
  card_token: "mwe42f64",
  stripe_response: response_body
)

• Search it with queries like: background_job.time_ms:>500
• Alert on it with threshold based alerts
• View this log's metadata in the console

...read more in our docs

The Timber.Context is local to each process, this is by design as it prevents processes from conflicting with each other as they maintain their contexts. But many times you'll want to share context between processes because they are related (such as processes created by Task or Flow). In these instances copying the context is easy.

current_context = Timber.LocalContext.get()

Task.async fn ->
  Timber.LocalContext.put(current_context)
  Logger.info("Logs from a separate process")
end

current_context in the above example is captured in the parent process, and because Elixir's variable scope is lexical, you can pass the referenced context into the newly created process. Timber.LocalContext.put/1 copies that context into the new process dictionary.

• Search it with queries like: background_job.time_ms:>500
• Alert on it with threshold based alerts
• View this log's metadata in the console

...read more in our docs

Capturing user context is a powerful feature that allows you to associate logs with users in your application. This is great for support as you can quickly narrow logs to a specific user, making it easy to identify user reported issues.

Simply add the UserContext immediately after you authenticate the user:

%Timber.Contexts.UserContext{id: "my_user_id", name: "John Doe", email: "[email protected]"}
|> Timber.add_context()

All of the UserContext attributes are optional, but at least one much be supplied.

Logging SQL queries can be useful but noisy. To reduce the volume of SQL queries you can limit your logging to queries that surpass an execution time threshold:

config :timber, Timber.Integrations.EctoLogger,
  query_time_ms_threshold: 2_000 # 2 seconds

In the above example, only queries that exceed 2 seconds in execution time will be logged.