Skip to main content

snowflake_put output

Sends messages to Snowflake stages and, optionally, calls Snowpipe to load this data into one or more tables.

# Common config fields, showing default values
output:
  label: ""
  snowflake_put:
    account: "" # No default (required)
    region: "" # No default (optional)
    cloud: "" # No default (optional)
    user: "" # No default (required)
    password: "" # No default (optional)
    private_key: "" # No default (optional)
    private_key_file: "" # No default (optional)
    private_key_pass: "" # No default (optional)
    role: "" # No default (required)
    database: "" # No default (required)
    warehouse: "" # No default (required)
    schema: "" # No default (required)
    stage: "" # No default (required)
    path: ""
    file_name: ""
    file_extension: ""
    compression: "AUTO"
    request_id: ""
    snowpipe: "" # No default (optional)
    batching:
      count: 0
      byte_size: 0
      period: ""
      check: ""
    max_in_flight: 1

In order to use a different stage and / or Snowpipe for each message, you can use function interpolations as described in Bloblang queries. When using batching, messages are grouped by the calculated stage and Snowpipe and are streamed to individual files in their corresponding stage and, optionally, a Snowpipe insertFiles REST API call will be made for each individual file.

Credentials​

Two authentication mechanisms are supported:

  • User/password
  • Key Pair Authentication

User/password​

This is a basic authentication mechanism which allows you to PUT data into a stage. However, it is not compatible with Snowpipe.

Key pair authentication​

This authentication mechanism allows Snowpipe functionality, but it does require configuring an SSH Private Key beforehand. Please consult the documentation for details on how to set it up and assign the Public Key to your user.

Note that the Snowflake documentation used to suggest using this command:

openssl genrsa 2048 | openssl pkcs8 -topk8 -inform PEM -out rsa_key.p8

to generate an encrypted SSH private key. However, in this case, it uses an encryption algorithm called pbeWithMD5AndDES-CBC, which is part of the PKCS#5 v1.5 and is considered insecure. Due to this, Expanso Edge does not support it and, if you wish to use password-protected keys directly, you must use PKCS#5 v2.0 to encrypt them by using the following command (as the current Snowflake docs suggest):

openssl genrsa 2048 | openssl pkcs8 -topk8 -v2 des3 -inform PEM -out rsa_key.p8

If you have an existing key encrypted with PKCS#5 v1.5, you can re-encrypt it with PKCS#5 v2.0 using this command:

openssl pkcs8 -in rsa_key_original.p8 -topk8 -v2 des3 -out rsa_key.p8

Please consult the pkcs8 command documentation for details on PKCS#5 algorithms.

Batching​

It's common to want to upload messages to Snowflake as batched archives. The easiest way to do this is to batch your messages at the output level and join the batch of messages with an archive and/or compress processor.

For the optimal batch size, please consult the Snowflake documentation.

Snowpipe​

Given a table called EVENTS_TBL with one column of type variant:

CREATE OR REPLACE TABLE EVENTS_DB.PUBLIC.EVENTS_TBL(RECORD variant)

and the following EVENTS_PIPE Snowpipe:

CREATE OR REPLACE PIPE EVENTS_DB.PUBLIC.EVENTS_PIPE AUTO_INGEST = FALSE AS COPY INTO EVENTS_DB.PUBLIC.EVENTS_TBL FROM (SELECT * FROM @%EVENTS_TBL) FILE_FORMAT = (TYPE = JSON COMPRESSION = AUTO)

you can configure Expanso Edge to use the implicit table stage @%EVENTS_TBL as the stage and EVENTS_PIPE as the snowpipe. In this case, you must set compression to AUTO and, if using message batching, you'll need to configure an archive processor with the concatenate format. Since the compression is set to AUTO, the gosnowflake client library will compress the messages automatically so you don't need to add a compress processor for message batches.

If you add STRIP_OUTER_ARRAY = TRUE in your Snowpipe FILE_FORMAT definition, then you must use json_array instead of concatenate as the archive processor format.

NOTE: Only Snowpipes with FILE_FORMAT TYPE JSON are currently supported.

Snowpipe troubleshooting​

Snowpipe provides the insertReport and loadHistoryScan REST API endpoints which can be used to get information about recent Snowpipe calls. In order to query them, you'll first need to generate a valid JWT token for your Snowflake account. There are two methods for doing so:

snowsql --private-key-path rsa_key.p8 --generate-jwt -a <account> -u <user>
  • Using the Python sql-api-generate-jwt utility:
python3 sql-api-generate-jwt.py --private_key_file_path=rsa_key.p8 --account=<account> --user=<user>

Once you successfully generate a JWT token and store it into the JWT_TOKEN environment variable, then you can, for example, query the insertReport endpoint using curl:

curl -H "Authorization: Bearer ${JWT_TOKEN}" "https://<account>.snowflakecomputing.com/v1/data/pipes/<database>.<schema>.<snowpipe>/insertReport"

If you need to pass in a valid requestId to any of these Snowpipe REST API endpoints, you can set a uuid_v4() string in a metadata field called request_id, log it via the log processor and then configure request_id: ${ @request_id } ). Alternatively, you can enable debug logging and Expanso Edge will print the Request IDs that it sends to Snowpipe.

General troubleshooting​

The underlying gosnowflake driver requires write access to the default directory to use for temporary files. Please consult the os.TempDir docs for details on how to change this directory via environment variables.

A silent failure can occur due to this issue, where the underlying gosnowflake driver doesn't return an error and doesn't log a failure if it can't figure out the current username. One way to trigger this behavior is by running Expanso Edge in a Docker container with a non-existent user ID (such as --user 1000:1000).

Performance​

This output benefits from sending multiple messages in flight in parallel for improved performance. You can tune the max number of in flight messages (or message batches) with the field max_in_flight.

This output benefits from sending messages as a batch for improved performance. Batches can be formed at both the input and output level. You can find out more in this doc.

Examples​

Kafka / realtime brokers​

Upload message batches from realtime brokers such as Kafka persisting the batch partition and offsets in the stage path and filename similarly to the Kafka Connector scheme and call Snowpipe to load them into a table. When batching is configured at the input level, it is done per-partition.

input:
  redpanda:
    seed_brokers:
      - localhost:9092
    topics:
      - foo
    consumer_group: rpcn
    max_yield_batch_bytes: 8MB
  processors:
    - mapping: |
        meta kafka_start_offset = meta("kafka_offset").from(0)
        meta kafka_end_offset = meta("kafka_offset").from(-1)
        meta batch_timestamp = if batch_index() == 0 { now() }
    - mapping: |
        meta batch_timestamp = if batch_index() != 0 { meta("batch_timestamp").from(0) }

output:
  snowflake_put:
    account: my-account
    user: [email protected]
    private_key_file: path_to_ssh_key.pem
    role: ACCOUNTADMIN
    database: EVENTS_DB
    warehouse: COMPUTE_WH
    schema: PUBLIC
    stage: "@%EVENTS_TBL"
    path: events/EVENTS_TBL/${! @kafka_partition }
    file_name: ${! @kafka_start_offset }_${! @kafka_end_offset }_${! meta("batch_timestamp") }
    upload_parallel_threads: 4
    compression: NONE
    snowpipe: EVENTS_PIPE

No compression​

Upload concatenated messages into a .json file to a table stage without calling Snowpipe.

output:
  snowflake_put:
    account: my-account
    user: [email protected]
    private_key_file: path_to_ssh_key.pem
    role: ACCOUNTADMIN
    database: EVENTS_DB
    warehouse: COMPUTE_WH
    schema: PUBLIC
    stage: "@%EVENTS_TBL"
    path: events
    upload_parallel_threads: 4
    compression: NONE
    batching:
      count: 10
      period: 3s
      processors:
        - archive:
            format: concatenate

Parquet format with snappy compression​

Upload concatenated messages into a .parquet file to a table stage without calling Snowpipe.

output:
  snowflake_put:
    account: my-account
    user: [email protected]
    private_key_file: path_to_ssh_key.pem
    role: ACCOUNTADMIN
    database: EVENTS_DB
    warehouse: COMPUTE_WH
    schema: PUBLIC
    stage: "@%EVENTS_TBL"
    path: events
    file_extension: parquet
    upload_parallel_threads: 4
    compression: NONE
    batching:
      count: 10
      period: 3s
      processors:
        - parquet_encode:
            schema:
              - name: ID
                type: INT64
              - name: CONTENT
                type: BYTE_ARRAY
            default_compression: snappy

Automatic compression​

Upload concatenated messages compressed automatically into a .gz archive file to a table stage without calling Snowpipe.

output:
  snowflake_put:
    account: my-account
    user: [email protected]
    private_key_file: path_to_ssh_key.pem
    role: ACCOUNTADMIN
    database: EVENTS_DB
    warehouse: COMPUTE_WH
    schema: PUBLIC
    stage: "@%EVENTS_TBL"
    path: events
    upload_parallel_threads: 4
    compression: AUTO
    batching:
      count: 10
      period: 3s
      processors:
        - archive:
            format: concatenate

DEFLATE compression​

Upload concatenated messages compressed into a .deflate archive file to a table stage and call Snowpipe to load them into a table.

output:
  snowflake_put:
    account: my-account
    user: [email protected]
    private_key_file: path_to_ssh_key.pem
    role: ACCOUNTADMIN
    database: EVENTS_DB
    warehouse: COMPUTE_WH
    schema: PUBLIC
    stage: "@%EVENTS_TBL"
    path: events
    upload_parallel_threads: 4
    compression: DEFLATE
    snowpipe: EVENTS_PIPE
    batching:
      count: 10
      period: 3s
      processors:
        - archive:
            format: concatenate
        - mapping: |
            root = content().compress("zlib")

RAW_DEFLATE compression​

Upload concatenated messages compressed into a .raw_deflate archive file to a table stage and call Snowpipe to load them into a table.

output:
  snowflake_put:
    account: my-account
    user: [email protected]
    private_key_file: path_to_ssh_key.pem
    role: ACCOUNTADMIN
    database: EVENTS_DB
    warehouse: COMPUTE_WH
    schema: PUBLIC
    stage: "@%EVENTS_TBL"
    path: events
    upload_parallel_threads: 4
    compression: RAW_DEFLATE
    snowpipe: EVENTS_PIPE
    batching:
      count: 10
      period: 3s
      processors:
        - archive:
            format: concatenate
        - mapping: |
            root = content().compress("flate")

Fields​

account​

Account name, which is the same as the Account Identifier. However, when using an Account Locator, the Account Identifier is formatted as <account_locator>.<region_id>.<cloud> and this field needs to be populated using the <account_locator> part.

Type: string

region​

Optional region field which needs to be populated when using an Account Locator and it must be set to the <region_id> part of the Account Identifier (<account_locator>.<region_id>.<cloud>).

Type: string

cloud​

Optional cloud platform field which needs to be populated when using an Account Locator and it must be set to the <cloud> part of the Account Identifier (<account_locator>.<region_id>.<cloud>).

Type: string

user​

Username.

Type: string

password​

An optional password.

Secret

This field contains sensitive information. Use a secret reference rather than a literal value.

Type: string

private_key​

The private SSH key. private_key_pass is required when using encrypted keys.

Secret

This field contains sensitive information. Use a secret reference rather than a literal value.

Type: string

private_key_file​

The path to a file containing the private SSH key. private_key_pass is required when using encrypted keys.

Type: string

private_key_pass​

An optional private SSH key passphrase.

Secret

This field contains sensitive information. Use a secret reference rather than a literal value.

Type: string

role​

Role.

Type: string

database​

Database.

Type: string

warehouse​

Warehouse.

Type: string

schema​

Schema.

Type: string

stage​

Stage name. Use either one of the supported stage types.

This field supports interpolation functions.

Type: string

path​

Stage path.

This field supports interpolation functions.

Type: string
Default: ""

file_name​

Stage file name. Will be equal to the Request ID if not set or empty.

This field supports interpolation functions.

Type: string
Default: ""

file_extension​

Stage file extension. Will be derived from the configured compression if not set or empty.

This field supports interpolation functions.

Type: string
Default: ""

upload_parallel_threads​

Specifies the number of threads to use for uploading files.

Type: int
Default: 4

compression​

Compression type.

Type: string
Default: "AUTO"

OptionSummary
AUTOCompression (gzip) is applied automatically by the output and messages must contain plain-text JSON. Default file_extension: gz.
DEFLATEMessages must be pre-compressed using the zlib algorithm (with zlib header, RFC1950). Default file_extension: deflate.
GZIPMessages must be pre-compressed using the gzip algorithm. Default file_extension: gz.
NONENo compression is applied and messages must contain plain-text JSON. Default file_extension: json.
RAW_DEFLATEMessages must be pre-compressed using the flate algorithm (without header, RFC1951). Default file_extension: raw_deflate.
ZSTDMessages must be pre-compressed using the Zstandard algorithm. Default file_extension: zst.

request_id​

Request ID. Will be assigned a random UUID (v4) string if not set or empty.

This field supports interpolation functions.

Type: string
Default: ""

snowpipe​

An optional Snowpipe name. Use the <snowpipe> part from <database>.<schema>.<snowpipe>. private_key or private_key_file must be set when using this feature.

This field supports interpolation functions.

Type: string

client_session_keep_alive​

Enable Snowflake keepalive mechanism to prevent the client session from expiring after 4 hours (error 390114).

Type: bool
Default: false

batching​

Allows you to configure a batching policy.

Type: object

batching.count​

A number of messages at which the batch should be flushed. If 0 disables count based batching.

Type: int
Default: 0

batching.byte_size​

An amount of bytes at which the batch should be flushed. If 0 disables size based batching.

Type: int
Default: 0

batching.period​

A period in which an incomplete batch should be flushed regardless of its size.

Type: string
Default: ""

batching.check​

A Bloblang query that should return a boolean value indicating whether a message should end a batch.

Type: string
Default: ""

batching.processors​

A list of processors to apply to a batch as it is flushed. This allows you to aggregate and archive the batch however you see fit. Please note that all resulting messages are flushed as a single batch, therefore splitting the batch into smaller batches using these processors is a no-op.

Type: array of processor

max_in_flight​

The maximum number of parallel message batches to have in flight at any given time.

Type: int
Default: 1