Function

Represents a deployed Modal Function, which can be invoked remotely.

class Function_ {
  readonly functionId: string;
  readonly methodName?: string;
  get logs(): FunctionLogsManager; // Access logs for this Function. Use `fetch()` to read logs from a UTC time range, `tail()` to read the most recent logs, and `stream()` to follow new logs as they arrive. See also: [`modal app logs`](https://modal.com/docs/cli/latest/app#modal-app-logs) for CLI access to logs for an App.
}

fromName

Accessed via modal.functions

async fromName(
  appName: string,
  name: string,
  params: FunctionFromNameParams = {},
): Promise<Function_>

Reference a Function by its name in an App.

Parameters (FunctionFromNameParams)

Optional parameters for client.functions.fromName().

• environment? (string)
• version? (number): Look up a version-pinned Function deployed at this App version.

getCurrentStats

async getCurrentStats(): Promise<FunctionStats>

getWebUrl

async getWebUrl(): Promise<string | undefined>

URL for addressing the Web Function via HTTP.

Returns: The web URL, or undefined if this is not a Web Function

instance

async instance(): Promise<Function_>

Create an instance of the Function with configuration specified by withOptions, withConcurrency, and/or withBatching.

remote

async remote(
  args: any[] = [],
  kwargs: Record<string, any> = {},
): Promise<any>

spawn

async spawn(
  args: any[] = [],
  kwargs: Record<string, any> = {},
): Promise<FunctionCall>

updateAutoscaler

async updateAutoscaler(
  params: FunctionUpdateAutoscalerParams,
): Promise<FunctionAutoscalerSettings>

Parameters (FunctionUpdateAutoscalerParams)

Optional parameters for Function_.updateAutoscaler().

• minContainers? (number)
• maxContainers? (number)
• bufferContainers? (number)
• scaledownWindowMs? (number)

withBatching

withBatching(params: FunctionWithBatchingParams): Function_

Override the static Function batching configuration at runtime.

Parameters (FunctionWithBatchingParams)

Configuration options for Function_.withBatching().

• maxBatchSize (number)
• waitMs (number)

withConcurrency

withConcurrency(params: FunctionWithConcurrencyParams): Function_

Override the static Function concurrency configuration at runtime.

Parameters (FunctionWithConcurrencyParams)

Configuration options for Function_.withConcurrency().

• maxInputs (number)
• targetInputs? (number)

withOptions

withOptions(options: FunctionWithOptionsParams): Function_

Override the static Function configuration at runtime.

Parameters (FunctionWithOptionsParams)

Configuration options for Function_.withOptions().

• cpu? (number)
• cpuLimit? (number)
• memoryMiB? (number)
• memoryLimitMiB? (number)
• gpu? (string)
• env? (Record<string, string>)
• secrets? (Secret[])
• volumes? (Record<string, Volume>)
• retries? (number | Retries)
• maxContainers? (number)
• bufferContainers? (number)
• scaledownWindowMs? (number)
• timeoutMs? (number)
• routingRegion? (string)

Function.logs

Access logs for this Function.

Use fetch() to read logs from a UTC time range, tail() to read the most recent logs, and stream() to follow new logs as they arrive.

See also: modal app logs for CLI access to logs for an App.

fetch

fetch(params: FunctionLogFetchParams): AsyncIterable<LogEntry>

Fetch Function logs corresponding to a UTC time range and optional filters.

Entries are returned in chronological order.

Parameters (FunctionLogFetchParams)

• since (Date): Start of the UTC time range.
• until? (Date): End of the UTC time range. Defaults to the current time.
• source? (LogSource): Filter by source: stdout, stderr, or system.
• searchText? (string): Filter by text contained in the log message.

Returns: An async iterable of LogEntry objects.

import { ModalClient } from "modal";

const modal = new ModalClient();
const function_ = await modal.functions.fromName("my-app", "train");

for await (const entry of function_.logs.fetch({
  since: new Date(Date.now() - 4 * 60 * 60 * 1_000),
  source: "stdout",
})) {
  process.stdout.write(entry.message);
}

stream

stream(params: FunctionLogStreamParams = {}): AsyncIterable<LogEntry>

Stream new Function logs until the timeout is reached.

Parameters (FunctionLogStreamParams)

• timeoutMs? (number): Number of milliseconds to wait between log entries before ending the stream. By default, the stream blocks until it is interrupted.

Returns: An async iterable of LogEntry objects as they arrive.

import { ModalClient } from "modal";

const modal = new ModalClient();
const function_ = await modal.functions.fromName("my-app", "train");

for await (const entry of function_.logs.stream({ timeoutMs: 60_000 })) {
  process.stdout.write(entry.message);
}

tail

tail(params: FunctionLogTailParams = {}): AsyncIterable<LogEntry>

Fetch the most recent Function logs.

Entries are returned in chronological order.

Parameters (FunctionLogTailParams)

• entries? (number): Number of log entries to return. Defaults to 100.
• source? (LogSource): Filter by source: stdout, stderr, or system.

Returns: An async iterable of LogEntry objects.

import { ModalClient } from "modal";

const modal = new ModalClient();
const function_ = await modal.functions.fromName("my-app", "train");

for await (const entry of function_.logs.tail({ entries: 20 })) {
  process.stdout.write(entry.message);
}