GCP Cloud Function in Python for Bolt

Sample Python Cloud Function in GCP that utilizes Cloud Storage Client library.

Requirements

• Python 3.0 or higher

Build From Source

• Download the source and change to the directory containing the sample code:

git clone https://gitlab.com/projectn-oss/projectn-bolt-python-gcf-sample.git 

cd projectn-bolt-python-gcf-sample

Deploy

To deploy the function, run the following command:

gcloud functions deploy <function-name> \
--entry-point bolt_gs_ops_handler \
--runtime python38 \
--trigger-http \
--service-account <service-account-email> \
--project <project-id> \
--region <region> \
--set-env-vars BOLT_URL=<Bolt-Service-Url>

Usage

The sample Python Cloud Function illustrates the usage and various operations, via separate entry points, that can be performed using Cloud Storage Client library for Python. The deployed Python Cloud Function can be tested from the Google Cloud Console by specifying a triggering event in JSON format.

Please ensure that Bolt is deployed before testing the sample Python Cloud Function. If you haven't deployed Bolt, follow the instructions given here to deploy Bolt.

Testing Bolt or GS Operations

bolt_gs_ops_handler is the function that enables the user to perform Bolt or GS operations. It sends a Bucket or Object request to Bolt or GS and returns an appropriate response based on the parameters passed in as input.

• bolt_gs_ops_handler represents a Google Cloud Function that is invoked by an HTTP Request.

• bolt_gs_ops_handler accepts the following input parameters as part of the HTTP Request:

  • sdkType - Endpoint to which request is sent. The following values are supported:

    • GS - The Request is sent to Google Cloud Storage.
    • Bolt - The Request is sent to Bolt, whose endpoint is configured via 'BOLT_URL' environment variable

  • requestType - type of request / operation to be performed. The following requests are supported:

    • list_objects - list objects
    • list_buckets - list buckets
    • get_object_md - head object
    • get_bucket_md - head bucket
    • download_object - get object (md5 hash)
    • upload_object - upload object
    • delete_object - delete object

  • bucket - bucket name

  • key - key name

• Following are examples of various HTTP requests, that can be used to invoke the function.

  • Listing objects from Bolt bucket:

      {"requestType": "list_objects", "sdkType": "BOLT", "bucket": "<bucket>"}

  • Listing buckets from GS:

    {"requestType": "list_buckets", "sdkType": "GS"}

  • Get Bolt object metadata (GET_OBJECT_MD):

    {"requestType": "get_object_md", "sdkType": "BOLT", "bucket": "<bucket>", "key": "<key>"}

  • Check if GS bucket exists (GET_BUCKET_MD):

    {"requestType": "get_bucket_md","sdkType": "GS", "bucket": "<bucket>"}

  • Download object (its MD5 Hash) from Bolt:

    {"requestType": "download_object", "sdkType": "BOLT", "bucket": "<bucket>", "key": "<key>"}

  • Upload object to Bolt:

    {"requestType": "upload_object", "sdkType": "BOLT", "bucket": "<bucket>", "key": "<key>", "value": "<value>"}

  • Delete object from Bolt:

    {"requestType": "delete_object", "sdkType": "BOLT", "bucket": "<bucket>", "key": "<key>"}

Data Validation Tests

bolt_gs_validate_obj_handler is the function that enables the user to perform data validation tests. It retrieves the object from Bolt and GS (Bucket Cleaning is disabled), computes and returns their corresponding MD5 hash. If the object is gzip encoded, object is decompressed before computing its MD5.

• bolt_gs_validate_obj_handler represents a Google Cloud Function that is invoked by an HTTP Request for performing data validation tests. To use this Function, change the entry point to bolt_gs_validate_obj_handler

• bolt_gs_validate_obj_handler accepts the following input parameters as part of the HTTP Request:

  • bucket - bucket name

  • key - key name

• Following is an example of a HTTP Request that can be used to invoke the function.

  • Retrieve object(its MD5 hash) from Bolt and GS:

    If the object is gzip encoded, object is decompressed before computing its MD5.

    {"bucket": "<bucket>", "key": "<key>"}

Performance Tests

bolt_gs_perf_handler is the function that enables the user to run Bolt or GS Performance tests. It measures the performance of Bolt or GS Operations and returns statistics based on the operation. Before using this handler, ensure that a source bucket has been crunched by Bolt with cleaner turned OFF. Get, List Objects tests are run using the first 1000 objects in the bucket and Put Object tests are run using objects of size 100 bytes. Delete Object tests are run on objects that were created by the Put Object test.

• bolt_gs_perf_handler represents a Google Cloud Function that is invoked by an HTTP Request for performing Bolt / GS Performance testing. To use this Function, change the entry point to bolt_gs_perf_handler.

• bolt_gs_perf_handler accepts the following input parameters as part of the HTTP Request:

  • requestType - type of request / operation to be performed. The following requests are supported:

    • list_objects - list objects
    • download_object - download object
    • download_object_ttfb - download object (first byte)
    • download_object_passthrough - download object (via passthrough) of unmonitored bucket
    • download_object_passthrough_ttfb - download object (first byte via passthrough) of unmonitored bucket
    • upload_object - upload object
    • delete_object - delete object
    • all - upload, download, delete, list objects (default request if none specified)

  • bucket - bucket name

• Following are examples of various HTTP requests, that can be used to invoke the function.

  • Measure List objects performance of Bolt / GS.

    {"requestType": "list_objects", "bucket": "<bucket>"}

  • Measure Download object performance of Bolt / GS.

    {"requestType": "download_object", "bucket": "<bucket>"}

  • Measure Download object (first byte) performance of Bolt / GS.

    {"requestType": "download_object_ttfb", "bucket": "<bucket>"} 

  • Measure Download object passthrough performance of Bolt.

    {"requestType": "download_object_passthrough", "bucket": "<unmonitored-bucket>"}

  • Measure Download object passthrough (first byte) performance of Bolt.

    {"requestType": "download_object_passthrough_ttfb", "bucket": "<unmonitored-bucket>"}

  • Measure Upload object performance of Bolt / GS.

    {"requestType": "upload_object", "bucket": "<bucket>"}

  • Measure Delete object performance of Bolt / GS.

    {"requestType": "delete_object", "bucket": "<bucket>"}

  • Measure Upload, Delete, Download, List objects performance of Bolt / GS.

    {"requestType": "all", "bucket": "<bucket>"}

Auto Heal Tests

bolt_auto_heal_handler is the function that enables the user to run auto heal tests. Before running this function, modify data-cruncher to use coldline tier-class and set backupduration and recoveryscannerperiod to 1 minute to ensure that the auto-healing duration is within the function execution timeout interval. Crunch a sample bucket having a single object. Then delete the single fragment object from the n-data bucket. Now run this function, passing the name of the crunched bucket along with the single object as input parameters to the function. The handler attempts to retrieve object repeatedly until it succeeds, which would indicate successful auto-healing of the object and returns the time taken to do so.

• bolt_auto_heal_handler represents a Google Cloud Function that is invoked by an HTTP Request for performing Auto-Heal testing. To use this Function, change the entry point to bolt_auto_heal_handler.

• BoltAutoHealHandler accepts the following input parameters as part of the event:

  • bucket - bucket name

  • key - key name

• Following is an example of a HTTP Request that can be used to invoke the function.

  • Measure Auto-Heal time of an object in Bolt.

    {"bucket": "<bucket>", "key": "<key>"}

Getting Help

For additional assistance, please refer to Project N Docs or contact us directly here