Monitor Netlify with Sazabi
Last verified on Nov 8, 2026.
Netlify serves your site from a CDN and runs your code in functions. Two different failures hide behind a broken page. A deploy can go wrong while it builds, or a function can go wrong while it serves. Netlify keeps these in separate places, and none of them is searchable next to the others.
Sazabi is AI-native observability and security. Through a Netlify log drain it receives the site's traffic, function, edge function and deploy logs as records you can search together, with the deploy ID tying a failing function to the build that shipped it.
What Sazabi reads from Netlify
The log drain is an HTTP push. Netlify batches log records from its CDN and build system and posts them in near real time. You choose the General HTTP endpoint service, paste your Sazabi intake URL and select JSON. The log drain requires a Netlify Enterprise plan, and you must be a Team Owner to create it (Netlify: Log Drains).
Netlify lists these data types: site traffic, function logs, edge function logs, deploy logs and Web Application Firewall rule matches. In the log drain form you pick the types you want, and you can exclude client_ip and user_agent from traffic logs if you do not want personal data to leave Netlify.
The fields differ by type:
- Function logs carry
function_name,function_type,duration,level,method,path,status_codeandrequest_id. - Edge function logs carry
function_name,request_path,levelandrequest_id, withfunction_typeset toedge. - Deploy logs carry
deploy_id,deploy_sectionandlevel. The section is one ofInitializing,Building,Deploying,CleanuporPost-Processing.
Every type also carries site_id, account_id and an RFC 3339 timestamp.
The other connection is the Netlify CLI. It puts the netlify command in the agent's sandbox, so the agent can inspect sites, deploys, functions and logs when you ask it in a thread. It does not feed data into the intake.
Questions it answers
- Which deploy introduced this function error, and which
deploy_sectiondid the failing build stop in? - Which functions have the longest
durationsince yesterday? - Are the
status_code502 responses concentrated in onefunction_name? - Did an edge function log an error for this
request_path? - Which deploys on a non-production branch failed while production stayed healthy?
A worked example: a function that keeps timing out
Say checkout requests are slow, and the site is on Enterprise with a log drain in place. Function logs include duration, which Netlify defines as the time AWS Lambda took to run the function. Ask for the slowest ones:
sazabi logs ask "Which Netlify functions had the highest duration in the last 24 hours? Show function_name, status_code and deploy_id"An example function record, with made-up values:
{
"function_name": "create-order",
"function_type": "regular",
"level": "REPORT",
"duration": 9873,
"method": "POST",
"path": "/.netlify/functions/create-order",
"status_code": 502,
"deploy_id": "61153ae8b0f6a900088386e8",
"request_id": "01FDWR77JMF2DA1CHF5YA6H07C",
"site_id": "example-site-id",
"timestamp": "2026-11-08T18:54:34.831Z"
}A REPORT line with a long duration and a 502 on one function names the suspect. Now ask whether the deploy_id is the newest deploy for the site, and read the deploy logs for it. If the function got slow right after a deploy, the build output or a changed dependency is the first place to look. If the deploy is old, look at the downstream call the function makes.
Limits to plan around
Netlify caps function log output at 4 KB per invocation, and if a log exceeds that only the last 4 KB is sent. A long stack trace may arrive cut off, so log the cause last. Netlify's docs also say the log drain does not support function log output for Background Functions, so those need another route to storage.
Sazabi expects JSON on this endpoint. If records look wrong, check the format in the log drain before anything else.
How it fits with other integrations
Netlify is one of several hosting platforms Sazabi reads. Compare it with Vercel and Render, or see all of them on the cloud platforms category page. If your site sits behind Cloudflare, the edge logs there explain traffic Netlify never saw. The full catalog lives on the integrations page.
Log sources
- Netlify (manual log source, cloud platforms): Configure a Netlify log drain pointed at a Sazabi endpoint to forward build, function, edge, and traffic logs in near real time.
What Sazabi detects from these logs today
These detections are on for every production project, with no monitors or alert rules to write. Sazabi opens an issue when one fires.
- Novel event: A kind of error that has never appeared before.
Frequently Asked Questions
Do I need a Netlify Enterprise plan to send logs to Sazabi?
Yes. Netlify Log Drains are available on Enterprise plans, and the Sazabi log source uses a Netlify log drain. If the Log Drains option is missing under Logs & Metrics, the site is on a lower tier. The Netlify CLI connection for the Sazabi agent does not use drains.
Which format should I choose for the Netlify log drain?
JSON. Choose the General HTTP endpoint service, paste your Sazabi intake URL as the full URL and set the format to JSON. Sazabi does not support other formats on this endpoint.
Which Netlify logs reach Sazabi?
The Sazabi docs describe build, function, edge and traffic logs. Netlify's own list also includes deploy logs and Web Application Firewall rule matches, and you pick the log types in the log drain form.
Can the Sazabi agent run Netlify commands?
Yes. Connect the Netlify CLI as a sandbox CLI with a personal access token in NETLIFY_AUTH_TOKEN, and the agent can run the netlify command to inspect sites, deploys, functions and logs from an investigation thread.
Sources
- Netlify log drain with the General HTTP endpoint service, JSON format, Enterprise plan requirement, build, function, edge and traffic logs (accessed Nov 8, 2026)
- Netlify CLI in the agent sandbox with NETLIFY_AUTH_TOKEN, to inspect sites, deploys, functions and logs (accessed Nov 8, 2026)
- Netlify has a manual log source and a sandbox CLI, and each method is configured separately (accessed Nov 8, 2026)
- Log log drain data types, function and deploy log fields (function_name, status_code, deploy_section), 4 KB function output limit, Background Functions not supported, Enterprise availability, PII exclusion option (accessed Nov 8, 2026)