System Audit Logs

Tecton's System Audit Logging allows you to see the configuration changes occurring in your Tecton account. System Audit Logging differs from Feature Request Audit Logs. The former records access-related changes in your Tecton account, while the latter records what requests are being sent to your production feature serving endpoint.

Accessing System Audit Logs

System Audit logs are available in your Cloud Provider's Object Storage that you configured for use with Tecton. For example, AWS customers will find their System Audit Logs in: s3://tecton-{DEPLOYMENT_NAME}/logging/system_audit_logs.

Note that Tecton on Snowflake customers do not have object storage configured, and must contact Support to access System Audit Logs.

Files are partitioned per day, and each filename is named with the ISO 8601 UTC timestamp that started the logging period. Filenames are formatted like YYYY-MM-dd/yyyyMMddTHHmmssZ.jsonl. Logs files are emitted in 15 minute intervals. Log events are available within 30 minutes of the event time.

For example, logs for the period starting at 08:00 UTC (inclusive) and ending at 08:15 UTC (exclusive) on July 1st 2023 can be found at

tecton-{DEPLOYMENT_NAME}/logging/system_audit_logs/2023-07-01/20230701T080000Z.jsonl

info

If no auditable events occur in a 15-minute interval, no file will be written for that interval.

Understanding System Audit Logs

System System Audit Logs are emitted as files of newline-separated json objects, with each object representing one auditable action.

Event Schema

Name	Description
timestamp	Timestamp of Event
actor	Actor Object id : Service Account or User ID type : USER, SERVICE_ACCOUNT, or TECTON_EMPLOYEE email : User email, only set for USER Actors
user_agent	User Agent Header
request_id	Request ID
event_type	<EVENT_TYPE>.<VERSION_NUMBER> (see possible EVENT_TYPE values here)
account_name	Tecton Account Name
request	Request Object (See event types for fields)
response	Response Object (See event types for fields)
status	Request Status
error_message	Error message (on failure only)

Sample Log Event

{
  "request_id": "eb88d142a4cd2cf5c1cc111e1f7a422f",
  "timestamp": "2023-07-20T21:31:55.826993Z",
  "account_name": "account_name",
  "event_type": "create_service_account.v1",
  "user_agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/114.0.0.0 Safari/537.36",
  "actor": {
    "type": "USER",
    "id": "00u38dayyxAF2xQ6O358",
    "email": "alice@tecton.ai"
  },
  "status": "OK",
  "request": {
    "name": "service_account_name",
    "description": "Service Account for CI/CD"
  },
  "response": {
    "id": "dgeb493c4d684b9xxx31d3b8ac5c0b09",
    "name": "service_account_name",
    "description": "Service Account for CI/CD",
    "is_active": false
  }
}

Event Types

The following are the event types currently captured in the System Audit Log.

event_type	Request Fields	Response Fields
create_workspace	workspace_name workspace_capabilities
delete_workspace	workspace
create_service_account	name description	id name description is_active
update_service_account	id name description is_active	id name description is_active
delete_service_account	id
create_account_user	login_email
delete_account_user	okta_id
account_user_action	okta_id resend_activation_email unlock_user grant_admin revoke_admin
assign_roles	List of Assignment Objects with: resource_type resource_id role principal_type principal_id
unassign_roles	List of Assignment Objects with: resource_type resource_id role principal_type principal_id
assign_roles_put	resource_type resource_id roles principal_type principal_id
set_feature_server_config	count	currentCount availableCount desiredCount

Versioning

If a field is ever removed from an event type, the version number will increase. Fields may be added to an event type object without a change in version number.

info

Version stability for System Audit Log event types is not supported at this time - events may be versioned at any time.

Using Audit Logs

System Audit Log events can be ingested into a customer's SIEM system (e.g. Microsoft Sentinel, Splunk, Sumo Logic, IBM QRadar, Securonix) for monitoring and alerting.

Alternatively, you can download System Audit Log files for local processing.

Downloading Audit Logs

You can download the audit logs files programmatically with:

• A client such as boto3 or Google Cloud Storage Client API
• Using your cloud provider's CLI
  • For Example:
    • AWS s3 sync
    • Google Cloud gcloud

The following is an example for how to download all files from S3 for July 1st 2023 (UTC) into a local directory:

import boto3
import os

BUCKET_NAME = "tecton-<DEPLOYENT_NAME>"
# Prefix matching your time range of interest, e.g. events on July 1st 2023
PREFIX = "logging/system_audit_logs/2023-07-01"
s3 = boto3.resource("s3")  # assumes credentials & configuration are handled outside python (e.g. in the .aws directory)
local_dir = "<SOME_LOCAL_DIRECTORY>"

bucket = s3.Bucket(BUCKET_NAME)
for obj in bucket.objects.filter(Prefix=PREFIX):
    filename = obj.key.split("/")[-1]  # filename will be like yyyyMMddTHHmmssZ.jsonl
    bucket.download_file(obj.key, os.join(local_dir, filename))

Processing System Audit Logs Locally

Below is one example of how to process System Audit Log files using Python.

import os
import pandas as pd
import json

local_dir = "<SOME_LOCAL_DIRECTORY>"  # directory containing only System Audit Log .jsonl files

events_as_string = []
for filename in os.listdir(local_dir):
    with open(os.path.join(local_dir, filename), "r") as f:
        events_as_string.extend([line for line in f.readlines()])

events_as_json = [json.loads(event) for event in events_as_string]

# events_as_json is now a list of json objects which can be analyzed however you like. One option is using pandas dataframes:

df = pd.DataFrame(events_as_json)

# Find all events where a user was granted administrative access
f = (df["event_type"] == "account_user_action.v1") & df.apply(lambda x: x["request"].get("grant_admin", False), axis=1)
grant_admin_events_only = df[f]

# Find all events where the action was taken by alice@tecton.ai
g = df.apply(lambda x: x["actor"].get("email", "") == "alice@tecton.ai", axis=1)
alice_actions_only = df[g]