Collect Duo authentication logs

This document explains how to ingest Duo authentication logs to Google Security Operations using Amazon S3. The parser extracts the logs from JSON formatted messages. It transforms the raw log data into the Unified Data Model (UDM), mapping fields like user, device, application, location, and authentication details, while also handling various authentication factors and results to categorize security events. The parser also performs data cleaning, type conversion, and error handling to ensure data quality and consistency.

Before you begin

• Google SecOps instance
• Privileged access to Duo tenant (Admin API application)
• Privileged access to AWS (S3, IAM, Lambda, EventBridge)

Configure Duo Admin API application

1. Sign in to Duo Admin Panel.
2. Go to Applications > Protect an Application.
3. Add Admin API application.
4. Copy and save the following values to a secure location:
   • Integration key (ikey)
   • Secret key (skey)
   • API hostname (for example, api-XXXXXXXX.duosecurity.com)
5. In Permissions, enable Grant read log (to read authentication logs).
6. Save the application.

Configure AWS S3 bucket and IAM for Google SecOps

1. Create Amazon S3 bucket following this user guide: Creating a bucket
2. Save bucket Name and Region for future reference (for example, duo-auth-logs).
3. Create a user following this user guide: Creating an IAM user.
4. Select the created User.
5. Select the Security credentials tab.
6. Click Create Access Key in the Access Keys section.
7. Select Third-party service as the Use case.
8. Click Next.
9. Optional: add a description tag.
10. Click Create access key.
11. Click Download CSV file to save the Access Key and Secret Access Key for later use.
12. Click Done.
13. Select the Permissions tab.
14. Click Add permissions in the Permissions policies section.
15. Select Add permissions.
16. Select Attach policies directly
17. Search for and select the AmazonS3FullAccess policy.
18. Click Next.
19. Click Add permissions.

Configure the IAM policy and role for S3 uploads

1. Go to AWS console > IAM > Policies > Create policy > JSON tab.

2. Enter the following policy:

   {
     "Version": "2012-10-17",
     "Statement": [
       {
         "Sid": "AllowPutDuoAuthObjects",
         "Effect": "Allow",
         "Action": "s3:PutObject",
         "Resource": "arn:aws:s3:::duo-auth-logs/*"
       },
       {
         "Sid": "AllowGetStateObject",
         "Effect": "Allow",
         "Action": "s3:GetObject",
         "Resource": "arn:aws:s3:::duo-auth-logs/duo/auth/state.json"
       }
     ]
   }
   
   

   • Replace duo-auth-logs if you entered a different bucket name.

3. Click Next > Create policy.

4. Go to IAM > Roles > Create role > AWS service > Lambda.

5. Attach the newly created policy.

6. Name the role WriteDuoAuthToS3Role and click Create role.

Create the Lambda function

1. In the AWS Console, go to Lambda > Functions > Create function.
2. Click Author from scratch.

3. Provide the following configuration details:

   Setting	Value
   Name	duo_auth_to_s3
   Runtime	Python 3.13
   Architecture	x86_64
   Execution role	WriteDuoAuthToS3Role

4. After the function is created, open the Code tab, delete the stub and enter the following code (duo_auth_to_s3.py):

   #!/usr/bin/env python3
   # Lambda: Pull Duo Admin API v2 Authentication Logs to S3 (raw JSON pages)
   # Notes:
   # - Duo v2 requires mintime/maxtime in *milliseconds* (13-digit epoch).
   # - Pagination via metadata.next_offset ("<millis>,<txid>").
   # - We save state (mintime_ms) in ms to resume next run without gaps.
   
   import os, json, time, hmac, hashlib, base64, email.utils, urllib.parse
   from urllib.request import Request, urlopen
   from urllib.error import HTTPError, URLError
   import boto3
   
   DUO_IKEY = os.environ["DUO_IKEY"]
   DUO_SKEY = os.environ["DUO_SKEY"]
   DUO_API_HOSTNAME = os.environ["DUO_API_HOSTNAME"].strip()
   S3_BUCKET = os.environ["S3_BUCKET"]
   S3_PREFIX = os.environ.get("S3_PREFIX", "duo/auth/").strip("/")
   STATE_KEY = os.environ.get("STATE_KEY", "duo/auth/state.json")
   LIMIT = min(int(os.environ.get("LIMIT", "500")), 1000)  # default 100, max 1000
   
   s3 = boto3.client("s3")
   
   def _canon_params(params: dict) -> str:
       parts = []
       for k in sorted(params.keys()):
           v = params[k]
           if v is None:
               continue
           parts.append(f"{urllib.parse.quote(str(k), '~')}={urllib.parse.quote(str(v), '~')}")
       return "&".join(parts)
   
   def _sign(method: str, host: str, path: str, params: dict) -> dict:
       now = email.utils.formatdate()
       canon = "\n".join([now, method.upper(), host.lower(), path, _canon_params(params)])
       sig = hmac.new(DUO_SKEY.encode("utf-8"), canon.encode("utf-8"), hashlib.sha1).hexdigest()
       auth = base64.b64encode(f"{DUO_IKEY}:{sig}".encode()).decode()
       return {"Date": now, "Authorization": f"Basic {auth}"}
   
   def _http(method: str, path: str, params: dict, timeout: int = 60, max_retries: int = 5) -> dict:
       host = DUO_API_HOSTNAME
       assert host.startswith("api-") and host.endswith(".duosecurity.com"), \
           "DUO_API_HOSTNAME must be like api-XXXXXXXX.duosecurity.com"
       qs = _canon_params(params)
       url = f"https://{host}{path}" + (f"?{qs}" if qs else "")
   
       attempt, backoff = 0, 1.0
       while True:
           req = Request(url, method=method.upper())
           req.add_header("Accept", "application/json")
           for k, v in _sign(method, host, path, params).items():
               req.add_header(k, v)
           try:
               with urlopen(req, timeout=timeout) as r:
                   return json.loads(r.read().decode("utf-8"))
           except HTTPError as e:
               if (e.code == 429 or 500 <= e.code <= 599) and attempt < max_retries:
                   time.sleep(backoff); attempt += 1; backoff *= 2; continue
               raise
           except URLError:
               if attempt < max_retries:
                   time.sleep(backoff); attempt += 1; backoff *= 2; continue
               raise
   
   def _read_state_ms() -> int | None:
       try:
           obj = s3.get_object(Bucket=S3_BUCKET, Key=STATE_KEY)
           val = json.loads(obj["Body"].read()).get("mintime")
           if val is None:
               return None
           # Backward safety: if seconds were stored, convert to ms
           return int(val) * 1000 if len(str(int(val))) <= 10 else int(val)
       except Exception:
           return None
   
   def _write_state_ms(mintime_ms: int):
       body = json.dumps({"mintime": int(mintime_ms)}).encode("utf-8")
       s3.put_object(Bucket=S3_BUCKET, Key=STATE_KEY, Body=body, ContentType="application/json")
   
   def _write_page(payload: dict, when_epoch_s: int, page: int) -> str:
       key = f"{S3_PREFIX}/{time.strftime('%Y/%m/%d', time.gmtime(when_epoch_s))}/duo-auth-{page:05d}.json"
       s3.put_object(
           Bucket=S3_BUCKET,
           Key=key,
           Body=json.dumps(payload, separators=(",", ":")).encode("utf-8"),
           ContentType="application/json",
       )
       return key
   
   def fetch_and_store():
       now_s = int(time.time())
       # Duo recommends a ~2-minute delay buffer; use maxtime = now - 120 seconds (in ms)
       maxtime_ms = (now_s - 120) * 1000
       mintime_ms = _read_state_ms() or (maxtime_ms - 3600 * 1000)  # 1 hour on first run
   
       page = 0
       total = 0
       next_offset = None
   
       while True:
           params = {"mintime": mintime_ms, "maxtime": maxtime_ms, "limit": LIMIT}
           if next_offset:
               params["next_offset"] = next_offset
   
           data = _http("GET", "/admin/v2/logs/authentication", params)
           _write_page(data, maxtime_ms // 1000, page)
           page += 1
   
           resp = data.get("response")
           items = resp if isinstance(resp, list) else []
           total += len(items)
   
           meta = data.get("metadata") or {}
           next_offset = meta.get("next_offset")
           if not next_offset:
               break
   
       # Advance window to maxtime_ms for next run
       _write_state_ms(maxtime_ms)
       return {"ok": True, "pages": page, "events": total, "next_mintime_ms": maxtime_ms}
   
   def lambda_handler(event=None, context=None):
       return fetch_and_store()
   
   if __name__ == "__main__":
       print(lambda_handler())
   
   

5. Go to Configuration > Environment variables > Edit > Add new environment variable.

6. Enter the following environment variables provided, replacing with your values:

   Key	Example
   S3_BUCKET	duo-auth-logs
   S3_PREFIX	duo/auth/
   STATE_KEY	duo/auth/state.json
   DUO_IKEY	DIXYZ...
   DUO_SKEY	****************
   DUO_API_HOSTNAME	api-XXXXXXXX.duosecurity.com
   LIMIT	500

7. After the function is created, stay on its page (or open Lambda > Functions > your‑function).

8. Select the Configuration tab.

9. In the General configuration panel click Edit.

10. Change Timeout to 5 minutes (300 seconds) and click Save.

Create an EventBridge schedule

1. Go to Amazon EventBridge > Scheduler > Create schedule.
2. Provide the following configuration details:
   • Recurring schedule: Rate (1 hour).
   • Target: your Lambda function.
   • Name: duo-auth-1h.
3. Click Create schedule.

Optional: Create read-only IAM user & keys for Google SecOps

1. In the AWS Console, go to IAM > Users, then click Add users.
2. Provide the following configuration details:
   • User: Enter a unique name (for example, secops-reader)
   • Access type: Select Access key - Programmatic access
   • Click Create user.
3. Attach minimal read policy (custom): Users > select secops-reader > Permissions > Add permissions > Attach policies directly > Create policy

4. In the JSON editor, enter the following policy:

   {
     "Version": "2012-10-17",
     "Statement": [
       {
         "Effect": "Allow",
         "Action": ["s3:GetObject"],
         "Resource": "arn:aws:s3:::<your-bucket>/*"
       },
       {
         "Effect": "Allow",
         "Action": ["s3:ListBucket"],
         "Resource": "arn:aws:s3:::<your-bucket>"
       }
     ]
   }
   

5. Set the name to secops-reader-policy.

6. Go to Create policy > search/select > Next > Add permissions.

7. Go to Security credentials > Access keys > Create access key.

8. Download the CSV (these values are entered into the feed).

Configure a feed in Google SecOps to ingest Duo Authentication Logs

1. Go to SIEM Settings > Feeds.
2. Click + Add New Feed.
3. In the Feed name field, enter a name for the feed (for example, Duo Authentication Logs).
4. Select Amazon S3 V2 as the Source type.
5. Select Duo Auth as the Log type.
6. Click Next.
7. Specify values for the following input parameters:
   • S3 URI: s3://duo-auth-logs/duo/auth/
   • Source deletion options: Select the deletion option according to your preference.
   • Maximum File Age: Default 180 Days.
   • Access Key ID: User access key with access to the S3 bucket.
   • Secret Access Key: User secret key with access to the S3 bucket.
   • Asset namespace: the asset namespace.
   • Ingestion labels: the label applied to the events from this feed.
8. Click Next.
9. Review your new feed configuration in the Finalize screen, and then click Submit.

UDM Mapping Table

Need more help? Get answers from Community members and Google SecOps professionals.