On This Page

Home / Stream/ Integrations/ Sources/ Kafka/Amazon MSK Source

Amazon MSK Source

Cribl Stream supports receiving data records from an Amazon Managed Streaming for Apache Kafka (MSK) cluster. This Source automatically detects compressed data in Gzip, Snappy, or LZ4 format.

Type: Pull | TLS Support: YES | Event Breaker Support: No

Kafka uses a binary protocol over TCP. It does not support HTTP proxies, so Cribl Stream must receive events directly from senders. You might need to adjust your firewall rules to allow this traffic.

Configure Cribl Stream to Receive Data from Kafka Topics
​

  1. On the top bar, select Products, and then select Cribl Stream. Under Worker Groups, select a Worker Group. Next, you have two options:
    • To configure via QuickConnect, navigate to Routing > QuickConnect. Select Add Source and select the Source you want from the list, choosing either Select Existing or Add New.
    • To configure via the Routes, select Data > Sources. Select the Source you want. Next, select Add Source.
  2. In the New Source modal, configure the following under General Settings:
    • Input ID: Enter a unique name to identify this Source definition. If you clone this Source, Cribl Stream will add -CLONE to the original Input ID.
    • Description: Optionally, enter a description.
    • Bootstrap servers: List of Kafka bootstrap servers to connect to, for example, kafkaBrokerHost:9092. Specify hostname and port – for example, mykafkabroker:9092 – or just hostname, in which case Stream will assign port 9092.
    • Topics: Enter the name(s) of topics to subscribe to. Press Enter/Return between multiple entries.
    • Region: Select the name of the AWS Region where your Amazon MSK cluster is located.

To optimize performance and prevent excessive rebalancing, Cribl recommends subscribing each Kafka Source to only one topic. To subscribe to multiple topics, consider creating a dedicated Kafka Source for each one.

For the same reason, the Group ID (below) should be unique for each of your Kafka, Confluent Cloud, and Azure Event Hubs Sources. For details, see Controlling Rebalancing.

  1. Next, you can configure the following Optional Settings:
    • Group ID: The name of the consumer group to which this Cribl Stream instance belongs.
    • From beginning: Toggle this off if you want Cribl Stream, upon subscribing to a topic for the first time, to read new messages only. Otherwise Cribl Stream will read all messages available on the server, starting from the oldest.
    • Tags: Optionally, add tags that you can use for filtering and grouping in Cribl Stream’s UI. Use a tab or hard return between (arbitrary) tag names. These tags aren’t added to processed events.
  2. Optionally, you can adjust the TLS, Authentication, Assume Role, Processing, Retries, and Advanced settings, or Connected Destinations outlined in the sections below.
  3. Select Save, then Commit & Deploy.

TLS Settings (Client Side)
​

For Amazon MSK Sources and Destinations:

  • IAM is the only type of authentication that Cribl Stream supports.
  • Because IAM auth requires TLS, TLS is automatically enabled.

Validate server certs: Reject certificates that are not authorized by a CA in the CA certificate path, or by another trusted CA (such as the system’s CA). Defaults to toggled on.

Server name (SNI): Leave this field blank. See Connecting to Kafka below.

Certificate name: The name of the predefined certificate.

CA certificate path: Path on client containing CA certificates (in PEM format) to use to verify the server’s cert. Path can reference $ENV_VARS.

Private key path (mutual auth): Path on client containing the private key (in PEM format) to use. Path can reference $ENV_VARS. Use only if mutual auth is required.

Certificate path (mutual auth): Path on client containing certificates in (PEM format) to use. Path can reference $ENV_VARS. Use only if mutual auth is required.

Passphrase: Passphrase to use to decrypt private key.

Minimum TLS version: Optionally, select the minimum TLS version to accept from connections.

Maximum TLS version: Optionally, select the maximum TLS version to accept from connections.

Authentication
​

Use the Authentication method drop-down to select an AWS authentication method.

Auto: This default option uses the AWS SDK for JavaScript to automatically obtain credentials in the following order of attempts:

  1. IAM Roles for Amazon EC2: Loaded from AWS Identity and Access Management (IAM) roles attached to an EC2 instance.
  2. Shared Credentials File: Loaded from the shared credentials file (~/.aws/credentials).
  3. Environment Variables: Loaded from environment variables AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY.
  4. JSON File on Disk: Loaded from a JSON file on disk.
  5. Other Credential-Provider Classes: Other credential-provider classes provided by the AWS SDK for JavaScript.

The Auto method works both when running on AWS and in other environments where the necessary credentials are available through one of the above methods.

SSO Providers
​

When using the auto authentication method, you can leverage SSO providers like SAML and Okta to issue temporary credentials. These credentials should be set in the environment variables AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY. The AWS SDK will then use these environment variables to authenticate.

Manual: If not running on AWS, you can select this option to enter a static set of user-associated IAM credentials (your access key and secret key) directly or by reference. This is useful for Workers not in an AWS VPC, such as those running in a private cloud. The Manual option exposes these corresponding additional fields:

  • Access key: Enter your AWS access key. If not present, will fall back to the env.AWS_ACCESS_KEY_ID environment variable, or to the metadata endpoint for IAM role credentials.
  • Secret key: Enter your AWS secret key. If not present, will fall back to the env.AWS_SECRET_ACCESS_KEY environment variable, or to the metadata endpoint for IAM credentials.

Secret: If not running on AWS, you can select this option to supply a stored secret that references an AWS access key and secret key. The Secret option exposes this additional field:

  • Secret key pair: Use the drop-down to select an API key/secret key pair that you’ve configured in Cribl Stream’s secrets manager. A Create link is available to store a new, reusable secret.

Assume Role
​

When using Assume Role to access resources in a different region than Cribl Stream, you can target the AWS Security Token Service (STS) endpoint specific to that region by using the CRIBL_AWS_STS_REGION environment variable on your Worker Node. Setting an invalid region results in a fallback to the global STS endpoint.

Enable for MSK: Toggle on to use Assume Role credentials to access MSK.

AssumeRole ARN: Enter the Amazon Resource Name (ARN) of the role to assume.

External ID: Enter the External ID to use when assuming role. This is required only when assuming a role that requires this ID in order to delegate third-party access. For details, see AWS’ documentation.

Duration (seconds): Duration of the Assumed Role’s session, in seconds. Minimum is 900 (15 minutes). Maximum is 43200 (12 hours). Defaults to 3600 (1 hour).

Processing Settings
​

Fields
​

In this section, you can add Fields to each event using Eval-like functionality.

Name: Field name.

Value: JavaScript expression to compute field’s value, enclosed in quotes or backticks. (Can evaluate to a constant.)

Pre-Processing
​

In this section’s Pipeline drop-down list, you can select a single existing Pipeline or Pack to process data from this input before the data is sent through the Routes.

Retries
​

These settings provide flexibility in handling retries for failed messages, allowing you to balance between quick retries and avoiding excessive load on the system.

The default configuration starts with a 300 milliseconds retry interval, doubles the interval after each retry, and caps the maximum retry interval at 30 seconds. The system will attempt to retry the request up to five times before considering it a failure.

Max retries: Maximum number of times to retry a failed request before the message fails. Defaults to 5. Enter 0 to not retry at all. For example, if set to 5, the system will attempt to retry the request up to five times before considering it a failure.

Initial retry interval (ms): Initial value used to calculate the retry interval, in milliseconds. This value determines the starting point for the retry delay. The default (and minimum) is 300 ms (three seconds). The maximum allowed value is 600000 ms (10 minutes). For example, if set to 1000 ms, the first retry will occur after one second.

Backoff multiplier: Set the backoff multiplier (2-20) to control the retry frequency for failed messages. The multiplier is used in an exponential backoff formula to increase the delay between retries. For faster retries, use a lower multiplier. For slower retries with more delay between attempts, use a higher multiplier. For example, with an initial retry interval of 1000 ms and a multiplier of 2, the retry intervals will be 1,000 ms, 2,000 ms, 4,000 ms, and so on. See the Kafka documentation for details.

Backoff limit (ms): The maximum wait time for a retry, in milliseconds. This setting caps the exponential backoff delay to prevent excessively long wait times. The default (and minimum) value is 30000 ms (30 seconds), and the maximum is 180000 ms (180 seconds). For example, if the calculated retry interval exceeds 180,000 ms, the retry will occur after 180,000 ms instead.

Advanced Settings
​

Use these settings to fine-tune Cribl Stream’s integration with Kafka topics. If you are unfamiliar with these parameters, contact Cribl Support to understand the implications of changing the defaults.

Heartbeat interval (ms): Expected time between heartbeats to the consumer coordinator when using Kafka’s group management facilities. Value must be lower than sessionTimeout, and typically should not exceed 1/3 of the sessionTimeout value. Defaults to 3000 ms (3 seconds). For details, see the Kafka documentation.

Session timeout (ms): Timeout used to detect client failures when using Kafka’s group management facilities. If the client sends the broker no heartbeats before this timeout expires, the broker will remove this client from the group, and will initiate a rebalance. Value must be between the broker’s configured group.min.session.timeout.ms and group.max.session.timeout.ms. Defaults to 30000 ms (30 seconds). For details, see the Kafka documentation.

Rebalance timeout (ms): Maximum allowed time for each Worker to join the Group after a rebalance has begun. If the timeout is exceeded, the coordinator broker will remove the Worker from the Group. Defaults to 60000 ms (1 minute). For details, see the Kafka documentation.

Connection timeout (ms): Maximum time to wait for a connection to complete successfully. Defaults to 10000 ms (10 seconds). Valid range is 1000 to 3600000 ms (1 second to 1 hour). For details, see the Kafka documentation.

Request timeout (ms): Maximum time to wait for Kafka to respond to a request. Defaults to 60000 ms (1 minute). For details, see the Kafka documentation.

Authentication timeout (ms): Maximum time to wait for Kafka to respond to an authentication request. Defaults to 1000 (1 second).

Reauthentication threshold (ms): If the broker requires periodic reauthentication, this setting defines how long before the reauthentication timeout Cribl Stream initiates the reauthentication. Defaults to 10000 (10 seconds).

A small value for this setting, combined with high network latency, might prevent the Source from reauthenticating before the Kafka broker closes the connection.

A large value might cause the Source to send reauthentication messages too soon, wasting bandwidth.

The Kafka setting connections.max.reauth.ms controls the reuthentication threshold on the Kafka side.

Offset commit interval (ms): How often, in milliseconds, to commit offsets. If both this field and the Offset commit threshold are empty, Cribl Stream will commit offsets after each batch. If both fields are set, Cribl Stream will commit offsets when either condition is met.

Offset commit threshold: How many events are needed to trigger an offset commit. If both this field and the Offset commit interval are empty, Cribl Stream will commit offsets after each batch. If both fields are set, Cribl Stream will commit offsets when either condition is met.

Max bytes per partition: Maximum amount of data that Kafka will return per partition, per fetch request. Must equal or exceed the maximum message size (maxBytesPerPartition) that Kafka is configured to allow. Otherwise, Cribl Stream can get stuck trying to retrieve messages. Defaults to 1048576 (1 MB).

Max bytes: Maximum number of bytes that Kafka will return per fetch request. Defaults to 10485760 (10 MB).

Maximum number of errors per socket: Maximum number of consecutive request errors that can occur on a single socket connection before the connection is discarded and reestablished. This mitigates issues with idle connections, particularly those used for sending heartbeats, thereby reducing the number of rebalances in a consumer group. Default is 0, which disables this feature. Accepts values 1-100.

Environment: If you’re using GitOps, optionally use this field to specify a single Git branch on which to enable this configuration. If empty, the config will be enabled everywhere.

If you observe an excessive number of group rebalances, and/or you observe consumers not regularly pulling messages, try increasing the values of Heartbeat interval, Session timeout, and Rebalance timeout.

Connected Destinations
​

Select Send to Routes to enable conditional routing, filtering, and cloning of this Source’s data via the Routing table.

Select QuickConnect to send this Source’s data to one or more Destinations via independent, direct connections.

Connecting to Kafka
​

Leave the TLS Settings > Server name (SNI) field blank
​

In Cribl Stream’s Kafka-based Sources and Destinations (including this one), the Kafka library that Cribl Stream uses manages SNI (Server Name Indication) without any input from Cribl Stream. Therefore, you should leave the TLS Settings > Server name (SNI) field blank.

Setting this field in the Cribl Stream UI can cause traffic to be routed to the wrong brokers, because it interferes with the Kafka library’s operation.

Connecting to a Kafka cluster involves using hostnames for two key components: bootstrap servers and brokers.

Brokers are servers that comprise the storage layer in a Kafka cluster. Bootstrap servers handle the initial connection to the Kafka cluster, and then return the list of brokers. A broker list can run into the hundreds. Every Kafka cluster has a bootstrap.servers property, defined as either a single hostname:port K-V pair, or a list of them. If Cribl Stream tries to connect via one bootstrap server and that fails, Cribl Stream then tries another one on the list.

In the General Settings > Bootstrap servers list, you can enter either the hostnames of brokers that your Kafka server has been configured to use, or, the hostnames of one or more bootstrap servers. If Kafka returns a list of brokers that’s longer than the list you entered, Cribl Stream keeps the full list internally. Cribl Stream neither saves the list nor makes it available in the UI. The connection process simply starts at the beginning whenever the Source or Destination is started.

Here’s an overview of the connection process:

  1. From the General Settings > Bootstrap servers list – where each broker is listed as a hostname and port – Cribl Stream takes a hostname and resolves it to an IP address.

  2. Cribl Stream makes a connection to that IP address. Notwithstanding the fact that Cribl Stream resolved one particular hostname to that IP address, there may be many services running at that IP address – each with its own distinct hostname.

  3. Cribl Stream establishes TLS security for the connection.

Although SNI is managed by the Kafka library rather than in the Cribl Stream UI, you might want to know how it fits into the connection process. The purpose of the SNI is to specify one hostname – that is, service – among many that might be running on a given IP address within a Kafka cluster. Excluding the other services is one way that TLS makes the connection more secure.

Internal Fields
​

Cribl Stream uses a set of internal fields to assist in handling of data. These “meta” fields are not part of an event, but they are accessible, and Functions can use them to make processing decisions.

Fields for this Source:

  • __inputId
  • __topicIn (indicates the Kafka topic that the event came from; see __topicOut in our Kafka Destination documentation)
  • __partition
  • __schemaId (when using Schema Registry)
  • __key (when using Schema Registry)
  • __headers (when using Schema Registry)
  • __keySchemaIdIn (when using Schema Registry)
  • __valueSchemaIdIn (when using Schema Registry)

How Cribl Stream Pulls Data
​

Kafka treats all the Worker Nodes as members of a Consumer Group, and Kafka manages each Node’s data load. By default, Workers will poll every 5 seconds. In the case of Leader failure, Worker Nodes will continue to receive data as normal.

How Cribl Stream Handles the _time Field
​

Events that the Kafka Source emits always contain a _time field, and sometimes also an __origTime field. Here’s how Cribl Stream determines what to send:

  • If the incoming Kafka message contains no _time field, the message’s timestamp becomes the value of the emitted event’s _time field.
  • If the incoming Kafka message contains a _time field whose value is a timestamp in UNIX epoch time, that timestamp becomes the value of the emitted event’s _time field.
  • If the incoming Kafka message contains a _time field whose value is not a timestamp in UNIX epoch time (for example, an ISO or UTC timestamp), that becomes the value of the emitted event’s __origTime field, and the message’s timestamp becomes the value of the emitted event’s _time field.

Randomized Partition Assignment
​

In Cribl Stream v.4.8.2 and newer, partition assignment is randomized. The Worker Processes selected to consume data during partition assignment are selected randomly, reducing the possibility of uneven load distribution when there are fewer partitions than Worker Processes. Note that random assignment is not the same as balanced assignment, and even distribution of work is not guaranteed.

Controlling Rebalancing
​

When you configure multiple Sources that subscribe to different topics but all belong to the same consumer group, a state change affecting any Source in this consumer group will affect all the other Sources. Examples of state changes include: deploying new configs, adding or removing Worker Processes, and Worker Processes crashing.

Here’s an example – three Sources, three different topics, all in one consumer group:

  • Source_1 - Topic_1 - ConsumerGroup1
  • Source_2 - Topic_2 - ConsumerGroup1
  • Source_3 - Topic_3 - ConsumerGroup1

Imagine that Source 1 undergoes a state change event, such as a Worker Process crash. Source 2 and Source 3 will rebalance – stopping data flow until the rebalance completes.

Shared Worker Group Mitigation
​

If Sources that share a consumer group all deploy as part of the same Worker Group, changes will have smaller side effects than when Sources are spread across different Worker Groups. (Conversely, imagine a configuration where deploying new configs for Worker Group 1 caused rebalancing of topics in Worker Worker Group 2. This spillover would be especially undesirable.)

Bottom Line
​

Changes to any member of a consumer group affect all other members of that consumer group. To prevent this undesired behavior, make sure to use a unique Group ID for each Kafka, Confluent Cloud, Amazon MSK, and Azure Event Hubs Source.

Troubleshooting
​

The Source’s configuration modal has helpful tabs for troubleshooting:

Live Data: Try capturing live data to see real-time events as they are ingested. On the Live Data tab, click Start Capture to begin viewing real-time data.

Logs: Review and search the logs that provide detailed information about the ingestion process, including any errors or warnings that may have occurred.

You can also view the Monitoring page that provides a comprehensive overview of data volume and rate, helping you identify ingestion issues. Analyze the graphs showing events and bytes in/out over time.

Common Issues
​

KafkaJSProtocolError: Not authorized to access topics: [Topic authorization failed]
​

The username does not have read permissions for the specified topic.