Published on
How do you host n8n in Brazil and pay in reais?
Deploy the official n8n image on a platform with servers in Brazil, with the /home/node/.n8n directory on persistent storage, a managed PostgreSQL database, and the webhook, encryption key, and timezone variables set. On Zenifra, the official template’s setup (Basic plan with 10 GB of persistent storage and a PostgreSQL 18 db-starter database with 3 instances of 10 GB) costs R$ 76.00 per month on the monthly contract or R$ 85.28 for a full month on hourly billing, in reais and with no IOF tax; this guide walks through the template and the manual setup and compares them with n8n Cloud and a VPS, using sources read on November 6, 2026.
The short answer: what you need and what it costs
To host n8n in Brazil and pay in reais, run the official n8nio/n8n image on a platform with servers in the country, with the /home/node/.n8n directory on persistent storage, a managed PostgreSQL database, and the webhook URL, encryption key, and timezone variables configured. On Zenifra, the official n8n template’s setup (Basic plan with 10 GB of persistent storage and a PostgreSQL 18 db-starter database with 3 instances of 10 GB) costs R$ 76.00 per month on the monthly contract, or R$ 85.28 for a full month on hourly billing, which is the template’s default, paid by PIX, boleto, or card, with no currency conversion and no IOF tax.
n8n is an automation tool with a visual editor: each workflow starts from a trigger (a schedule, a webhook, an incoming message) and chains calls to APIs, databases, and services. The Community Edition, the version you host yourself, is free for internal use under the n8n license and has no per-execution fee; what limits how many workflows run at the same time is the CPU, memory, and database you pay for. n8n Cloud, the version hosted by n8n itself, charges by tiers of monthly executions and, according to its pricing page, stores the data of hosted plans in Frankfurt, Germany.
If you receive webhooks from Brazilian services and call APIs in Brazil, server location matters. In the benchmark Zenifra published, with residential probes in three state capitals, a new HTTPS connection from São Paulo took 76 ms to reach the Zenifra API and 630 ms to reach Frankfurt, in a single measurement that works as an order of magnitude. Every webhook that arrives on a new connection and every call the workflow makes in sequence pays for that distance.
This guide follows n8n’s official hosting documentation and the definition of the official n8n template in the Console’s template catalog, read on November 6, 2026. Zenifra prices come from the public catalog, which lists values in Brazilian cents; n8n Cloud prices come from the official pricing page on the same day. Dollar amounts are converted at the PTAX selling rate of November 5, 2026 (R$ 5.4621), plus the 3.5% IOF tax on international card purchases (Decree 6,306/2007, art. 15-B, item VII, as amended by Decree 12,499/2025).
What n8n needs to run in production
n8n’s hosting documentation makes clear what separates a test from a production install. The first item is the database: by default n8n saves workflows, credentials, and execution history in a SQLite file at ~/.n8n/database.sqlite, and it also supports PostgreSQL configured through environment variables. The second is the /home/node/.n8n directory: in the Docker install it is mounted on a volume so data survives restarts, and the documentation recommends keeping it persistent even with PostgreSQL, because it holds the encryption key, instance logs, and source control files.
The third item is the encryption key. On first launch n8n generates a random key, saves it in ~/.n8n, and uses it to encrypt credentials before writing them to the database. You can set your own with N8N_ENCRYPTION_KEY; if the key changes or is lost, saved credentials can no longer be read. The fourth is the webhook URL: n8n builds that URL from N8N_PROTOCOL, N8N_HOST, and N8N_PORT, which does not work behind a reverse proxy, where n8n listens on port 5678 and the public reaches it on 443. In that case the URL has to be set manually in N8N_WEBHOOK_URL, so the editor shows the right address and external services receive the public one.
The fifth is the timezone. n8n’s default is America/New_York, and that is the timezone the Schedule node uses to decide when a scheduled workflow starts. With GENERIC_TIMEZONE=America/Sao_Paulo, a schedule for 8 a.m. fires at 8 a.m. Brasília time. The Docker install also mentions the TZ variable, which sets the system timezone and changes what scripts and commands such as date return.
Beyond those five, plan for history growth. By default n8n deletes old executions once they are older than 336 hours (14 days) or once the total passes 10,000, and it lets you skip saving successful executions, each node’s progress, or manual runs. With SQLite, the space used by deleted executions does not go back to the disk on its own: it is reused, and to free it you need to enable DB_SQLITE_VACUUM_ON_STARTUP or run a manual VACUUM.
| Requirement | What the n8n documentation says | How it looks on Zenifra |
|---|---|---|
| Database | SQLite by default at ~/.n8n/database.sqlite; PostgreSQL 16, 17, and 18 supported | PostgreSQL 18 db-starter with 3 instances and 10 GB, recommended by the template and created with it under Customize |
| Data directory | Volume mounted at /home/node/.n8n, recommended even with PostgreSQL | 10 GB of persistent storage at /home/node/.n8n, set in the template |
| Encryption key | Generated on first launch and saved in ~/.n8n, or set in N8N_ENCRYPTION_KEY | With the template, n8n generates the key and keeps it in the persistent directory; in the manual setup, set N8N_ENCRYPTION_KEY before the first launch |
| Webhook URL | Set in N8N_WEBHOOK_URL behind a reverse proxy; WEBHOOK_URL is the old name, still accepted | Add N8N_WEBHOOK_URL=https://<your-address>.clients.zenifra.com/ to the project variables |
| Timezone | Default America/New_York; GENERIC_TIMEZONE for schedule nodes | Add GENERIC_TIMEZONE=America/Sao_Paulo to the project variables |
| Port | n8n listens on 5678 | Port 5678 in the template, with HTTPS on the public address |
A detail about the database on n8n Cloud: n8n’s own documentation states that the Starter and Pro plans use SQLite, and that PostgreSQL is reserved for Enterprise scaling plans. SQLite is not a mistake; it only gets tight as execution volume grows. That is why Zenifra’s official template already declares a PostgreSQL database as a dependency, instead of keeping everything in the local file.
Managed PostgreSQL: version, plan, connection, and backup
n8n supports the two latest actively maintained PostgreSQL major versions, 17 and 18 (as of July 2026 in the documentation), plus 16 for compatibility, and it does not support PostgreSQL-compatible derivatives such as AlloyDB, CockroachDB, or YugabyteDB. Zenifra’s database catalog offers PostgreSQL 15, 16, 17, and 18, and the template recommends 18, which is on n8n’s list; if you change the version under Customize, avoid 15. The n8n documentation also warns that this range shifts every year, in November, when the PostgreSQL project retires its oldest version and releases a new one.
The template recommends db-starter with 3 instances and a 10 GB disk on each. The plan has 0.3 vCPU and 256 MB per instance and automatic backups every 24 hours with 7-day retention. With 3 instances, the primary handles writes and reads, the other two are replicas, and there is automatic failover: if the primary fails, a replica can be promoted. During a failover, open connections may drop; the n8n documentation describes how it checks the database connection periodically and rebuilds it on its own after a drop, without a restart. As volume grows, db-basic (1 vCPU and 1 GB per instance, backups every 12 hours) is the next step.
With the template, you do not type in connection data. It ships four variables with fixed values, DB_TYPE=postgresdb, DB_POSTGRESDB_DATABASE=app, DB_POSTGRESDB_USER=app, and DB_POSTGRESDB_SCHEMA=public, and leaves the host, port, and password to the database data: when you select Create a database with the project, the variables the author linked are filled in automatically, stay read-only, and keep sensitive values masked. In the manual setup, you add the same seven variables using the database connection card in the Console: a host in the postgresql-<id>.managed.zenifra.com format, a port set by the platform, user app, and database app.
DB_TYPE=postgresdb
DB_POSTGRESDB_HOST=postgresql-<id>.managed.zenifra.com
DB_POSTGRESDB_PORT=<port shown in the Console>
DB_POSTGRESDB_DATABASE=app
DB_POSTGRESDB_USER=app
DB_POSTGRESDB_PASSWORD=<database password>
DB_POSTGRESDB_SCHEMA=publicThe database password appears only at creation or after a rotation. Once created, the database and the application are independent: if you rotate the password, update DB_POSTGRESDB_PASSWORD in n8n, because the application is not updated on its own. If the database serves only n8n, you can turn public access off and use the organization’s private connections between applications; in that case, switch DB_POSTGRESDB_HOST to the internal address before turning it off.
Automatic backups and replicas do not replace a copy of your own. Replicas protect against an instance failing, not against accidental deletion or corrupted data, and restoring a backup is requested through support, since there is no restore from the Console yet. The application’s persistent storage keeps three copies, but a file the application deletes is gone from all of them. For critical workflows, keep a periodic export outside the platform and remember that the encryption key lives in the config file under /home/node/.n8n: without it, the credentials saved in the database cannot be read.
Step by step with the n8n template
n8n is one of the Official Templates in Zenifra’s public catalog, published by Zenifra itself and marked as requiring a database. According to its definition in the Console’s template catalog, read on November 6, 2026, it deploys the docker.io/n8nio/n8n:2.32.3 image on port 5678, with 10 GB of persistent storage at /home/node/.n8n, 1 instance, the Basic plan, and hourly billing as defaults, and it recommends a PostgreSQL 18 db-starter database with 3 instances and 10 GB. To use it you need the project.create permission in the organization and, to create the database along with it, database.create; the organization owner has both.
There are two paths. Deploy creates the project with all of the template’s recommendations, after a confirmation that shows the automatic name (the template name with a random eight-character suffix) and a summary of plan, billing, access, instances, persistence, and auto-scaling. Customize opens project creation with everything filled in and lets you change the name, description, plan, billing mode, access, instances, capacity, and variable values. The application, the port, the HTTP type, the persistent folder, and the set of variables stay fixed: the form accepts exactly the declared variables.
For n8n, choose Customize for two reasons. First, only Customize offers Create a database with the project: quick deployment through Deploy creates only the application, and n8n would be left without the PostgreSQL database its variables expect. Second, that is where you change the billing mode: the default is hourly, good for testing, and for an n8n instance that stays on all month the monthly contract is cheaper. The steps below follow the Templates documentation.
- 1. In the Console menu, open Templates.
- 2. Under Public templates, use Official or Most used and open the n8n card.
- 3. Review the author, tags, revision, application, port, persistence, variables, and defaults.
- 4. Click Customize and adjust the project name, the plan (the default is Basic), and the billing mode (the default is hourly).
- 5. Select Create a database with the project and check the recommended configuration: PostgreSQL 18, db-starter, 3 instances, and 10 GB.
- 6. Review the summary of both resources and their costs, then click Create project.
- 7. After creation, open the n8n project under Edit project and, in Environment variables, add
N8N_WEBHOOK_URLwith the public address andGENERIC_TIMEZONE=America/Sao_Paulo. Saving restarts the instances.
Step seven is our recommendation, based on the n8n documentation: the template does not define these variables, and without N8N_WEBHOOK_URL the editor has no way to know the public address it should show in the Webhook node. Do not add N8N_ENCRYPTION_KEY after n8n has already started: on first launch it generates the key and writes it to the persistent directory, and n8n’s code refuses to start when the variable holds a key different from the stored one (a mismatching keys error). The generated key stays in the project’s persistent storage.
If any creation step fails, Zenifra rolls back every resource created in that operation, and the template does not gain a new use. After success, the application and the database become independent projects: deleting one does not delete the other. The project keeps a reference to the template revision it used, but future template changes do not alter your project; updating the n8n version remains your decision. The Official Template badge identifies the publisher’s origin, not a guarantee for every scenario, so check that the plan fits the workflow volume you expect.
Step by step with the manual setup
If you prefer to build the project from scratch, use the same values as the template in an HTTP project with the official Docker Hub image. Create the PostgreSQL 18 db-starter database with 3 instances and 10 GB first, so you have the host, port, and password at hand. Then, in the Console, click + Create Project, choose HTTP, and fill in the fields in the table. The name must be 6 to 32 characters long, with lowercase letters, numbers, and hyphens, and it defines the *.clients.zenifra.com address; a custom subdomain, of 3 to 54 characters, starts on the Premium plan.
Persistent storage is required. With PostgreSQL, workflows and executions live in the database, but the encryption key, instance logs, and other n8n files live in /home/node/.n8n; without persistent storage, that content is lost on every restart. Set it carefully at creation, because the persisted directory cannot be changed later. Capacity ranges from 1 to 250 GB; with hourly billing it can be increased at any time, never reduced, and on monthly or yearly billing it stays fixed for the contract.
In the Environment variables step, add the seven database variables, the same as the template’s, shown in the previous section. As our recommendation based on the n8n documentation, add three more: N8N_WEBHOOK_URL with the public address, N8N_ENCRYPTION_KEY with a long random string generated by a password manager, and GENERIC_TIMEZONE=America/Sao_Paulo. Set the key before the first launch and never change it afterwards, because n8n refuses to start with a key different from the one it stored; keep a copy in a password vault. WEBHOOK_URL is the old name of the webhook variable and is still accepted, with a deprecation warning in the log.
Review the estimated cost and click Create Project; when it finishes, the Console opens the project page with the public address. On first access, n8n asks you to create the owner account: create it right away, before sharing the address. Then create a workflow with a Webhook node and confirm that the URL shown uses the address in N8N_WEBHOOK_URL. The node has two URLs: the test URL, which works while you listen for a test event and shows the data in the editor, and the production URL, which n8n registers when you publish the workflow. If the application does not respond, check the project logs and the configured port.
# Recommended extra variables (n8n documentation)
N8N_WEBHOOK_URL=https://<your-address>.clients.zenifra.com/
N8N_ENCRYPTION_KEY=<long-random-key>
GENERIC_TIMEZONE=America/Sao_Paulo| Field | Value, same as the official template |
|---|---|
| Project source | OCI Image |
| Image | docker.io/n8nio/n8n:2.32.3 |
| Registry | Public, no authentication |
| Plan | Basic (0.5 vCPU and 512 MB), the template default; Premium or Premium Plus for more volume |
| Billing | Hourly, the template default; monthly for an n8n instance that stays on all month |
| Port | 5678 |
| Instances | 1 |
| Persistent storage | Yes, 10 GB, directory /home/node/.n8n |
About the version: the pinned 2.32.3 tag avoids surprise upgrades, which is the right call in production. On November 6, 2026, the n8n documentation lists 2.46.6 as the stable version and says a new minor version ships most weeks; to upgrade, change the image tag, read the release notes, and watch the logs on the first startup.
The maximum webhook payload size is 16 MB by default; for self-hosted installs, the Webhook node documentation points to the N8N_PAYLOAD_SIZE_MAX variable to change that limit. To remove the project, open it under Projects and click Delete. Deletion is permanent and erases storage, so export your workflows first.
What it costs per month on Zenifra, line by line
The calculation below adds up the four items of the template’s setup: the Basic application plan, 10 GB of persistent storage, a db-starter PostgreSQL database with 3 instances, and a 10 GB disk on each instance. Values come from Zenifra’s public catalog, which lists prices in Brazilian cents, checked on November 6, 2026. The reference month has 720 hours (30 days), the same basis as the storage catalog, where the hourly price is the monthly price divided by 720.
Persistent storage for an HTTP application is replicated in three copies and costs R$ 0.44 per GB per month. On PostgreSQL, both the plan and the disk are billed per instance: the disk costs R$ 0.20 per GB per month on each one, so the template’s 10 GB becomes 30 GB billed across the 3 instances. The 5 GB shown in the Basic plan description is ephemeral storage, so the 10 GB of persistent storage is billed in full. On the monthly contract, storage is part of the contract and charged at creation; on the yearly one, at 12 times the monthly price per GB.
- Template (Basic, 10 GB persistent, db-starter with 3 instances of 10 GB): R$ 85.28 for a full month on hourly billing, R$ 76.00 monthly, R$ 846.40 yearly.
- More CPU and metrics (Premium with 1 vCPU and 1 GB, the same 10 GB and the same database): R$ 128.48 for a full month on hourly billing, R$ 109.00 monthly, R$ 1,175.40 yearly.
- High volume and IP access control (Premium Plus with 2 vCPU and 2 GB, the same 10 GB, db-basic with 3 instances of 10 GB): R$ 332.96 for a full month on hourly billing, R$ 280.00 monthly, R$ 3,087.40 yearly.
| Item | Catalog (cents) | Hourly, 720 h | Monthly | Yearly |
|---|---|---|---|---|
| Basic application (0.5 vCPU, 512 MB) | 5 per hour; 2,990 per month; 32,890 per year | R$ 36.00 | R$ 29.90 | R$ 328.90 |
| Application persistent storage, 10 GB | 44 per GB-month | R$ 4.40 | R$ 4.40 | R$ 52.80 |
| db-starter PostgreSQL, 3 instances | 1.8 per hour; 1,190 per month; 13,090 per year, per instance | R$ 38.88 | R$ 35.70 | R$ 392.70 |
| Database disk, 10 GB on each of the 3 instances (30 GB) | 20 per GB-month | R$ 6.00 | R$ 6.00 | R$ 72.00 |
| Total | Sum of the four items | R$ 85.28 per month | R$ 76.00 per month | R$ 846.40 per year |
With hourly billing, Zenifra adds up the project’s running hours in the cycle, multiplied by the number of instances, and rounds the total up to the next full hour; a stopped project is not charged for compute, but persistent storage is still billed. Hourly is the template’s default and makes sense for testing: with n8n on for the whole month, the monthly contract is R$ 9.28 cheaper. Plan changes are only available on hourly billing; on monthly and yearly contracts, the plan, instances, and storage stay fixed for the contract.
Basic covers the start, but it has no CPU, memory, and request metrics, no healthcheck, and no custom subdomain, which start on Premium. If your workflows grow, or if you want to watch memory use before the application restarts from running out of it, the scenarios above show what moving up a plan costs. The organization pays by PIX, boleto, or card, always in reais, and credit added by PIX or boleto is used before the card; in the site’s calculator, set the database to 3 instances and 10 GB to reproduce the math.
n8n Cloud: the official price and what you get for it
n8n Cloud is the version hosted and run by n8n itself. The pricing page, opened from Brazil on November 6, 2026, shows prices in reais: Starter at R$ 150 per month on the monthly plan or R$ 1,500 per year on the yearly plan, and Pro at R$ 375 per month on the monthly plan or R$ 3,750 per year on the yearly plan. Business, at R$ 5,000 per month on the monthly plan, is currently sold only for self-hosted installs. Every plan includes unlimited users and workflows; what changes is the monthly execution quota, the number of concurrent executions, and the maximum duration of each execution.
For visitors from the United States, the price list the page fetches from Paddle, the payment platform used at checkout, shows US$ 24 per month for Starter and US$ 60 for Pro. Converted at the PTAX rate of November 5, 2026 with 3.5% IOF, that is R$ 135.68 and R$ 339.20: the price in reais shown to Brazilian visitors is higher than the converted dollar price. The page does not say whether taxes or IOF apply to the price in reais, so check your card statement. Starter and Pro are bought on the website; payment by invoice or wire transfer is only available for yearly Business and Enterprise, according to the page’s FAQ.
| Plan | Price on the page, opened from Brazil | US dollar price, converted with PTAX and IOF | Executions per month | Concurrent executions | Maximum duration per execution |
|---|---|---|---|---|---|
| n8n Cloud Starter | R$ 150 per month (monthly) or R$ 1,500 per year (R$ 125 per month) | US$ 24 = R$ 135.68 per month | 2,500 | 5 | 5 minutes |
| n8n Cloud Pro, 10k tier | R$ 375 per month (monthly) or R$ 3,750 per year (the page shows R$ 313 per month) | US$ 60 = R$ 339.20 per month | 10,000 | 20 | 40 minutes |
| Zenifra: n8n template (Basic + db-starter with 3 instances) with the Community Edition | R$ 76.00 per month (monthly) or R$ 85.28 for a full month on hourly billing, in reais and with no IOF | Not applicable | No per-execution fee; the limit is the plan’s capacity | No plan quota | You set it in your install |
Where n8n Cloud wins: you do not run a server or a database, and the plans include Assistant credits (1,600 per month on Starter) and AI models without your own API key. The n8n documentation is direct about it: self-hosting is recommended for expert users, and people without server experience should choose n8n Cloud. If you run a few workflows, fit within the 2,500-execution quota, and have no one to look after versions and backups, Starter does the job.
Where self-hosting wins: there is no execution quota and no 5- or 40-minute per-execution limit imposed by a plan, your data stays in Brazil, and the template’s setup on Zenifra (R$ 76.00 per month on the monthly contract) costs R$ 74.00 less per month than Starter (R$ 150). The trade-off is that you take care of the n8n version, the backups, and the plan size. For companies with fewer than 20 employees, n8n advertises a Startup plan with 50% off Business, which is the license with SSO, environments, and version control for self-hosted installs.
A generic VPS: when it pays off and what stays with you
A VPS is a virtual machine where you install everything yourself. For n8n, that means installing Docker, running the image with the volume at /home/node/.n8n, setting up PostgreSQL or accepting SQLite, putting a reverse proxy with an HTTPS certificate in front, setting the webhook URL for that proxy, and handling the firewall, system updates, and backups. n8n’s Docker install documentation lists the skills you need without hedging: setting up and configuring servers and containers, managing resources and scaling, securing servers and applications, and configuring n8n itself, and it warns that mistakes can lead to data loss, security issues, and downtime.
A VPS pays off when you already run servers and want full control: running queue mode with Redis and several workers, installing custom nodes, tuning the operating system, or sharing the machine with other services. It can also be cheaper in raw CPU and memory. We do not compare provider prices here because each one has its own price list, and what usually weighs most is the time of whoever maintains the machine; the VPS or managed platform guide on this blog covers that split of responsibilities in more detail.
On Zenifra, HTTPS, CDN, and DDoS protection come with the HTTP plans, persistent storage is replicated in three copies, PostgreSQL includes automatic backups in the plan and failover with more than one instance and, with hourly billing, you change plans without recreating the project. In exchange, you accept the platform’s limits: the template uses a single instance for n8n, the persistent directory is set at creation, and queue mode is not part of the template.
| Task | On a VPS | On Zenifra |
|---|---|---|
| HTTPS and certificate | You install the reverse proxy and handle certificate renewal | HTTPS included on the *.clients.zenifra.com address and on custom domains |
| Webhook URL behind the proxy | You configure the proxy and the variable | N8N_WEBHOOK_URL pointing to the public address, in the project variables |
| Database | You install, update, replicate, and back up PostgreSQL | Managed PostgreSQL; in the template, 3 instances with failover and backups every 24 hours, 7-day retention |
| n8n disk | The machine’s disk; copies and space are up to you | Persistent storage with three copies, from 1 to 250 GB, with usage shown in the Console |
| Operating system | Updates, firewall, and SSH access are up to you | No operating system to manage; you look after the image and the variables |
| Scaling | Queue mode with Redis and workers, if you build it | Plan changes on hourly billing; a single instance for n8n, as in the template |
| Payment | Depends on the provider | In reais, by PIX, boleto, or card, with no IOF |
Latency: why it matters for webhooks from Brazilian services
A webhook is an HTTP request that an external service sends to your n8n when something happens: a payment approved, a message received, a form submitted. Every notification that arrives on a new connection pays for the TCP handshake, the TLS negotiation, and the wait for the first byte. The farther n8n is from the caller, the longer each notification takes, and the same applies in the other direction, to every call the workflow makes to APIs in Brazil.
In the benchmark Zenifra published, measured on November 6, 2026 with Globalping and residential probes in São Paulo, Rio de Janeiro, and Fortaleza, Zenifra’s public API answered a new HTTPS connection in 76 ms from São Paulo, 77 ms from Rio, and 144 ms from Fortaleza, as medians of 8 rounds and with real processing on the route. A reference endpoint in Washington took 378, 335, and 304 ms. Frankfurt, where n8n’s pricing page says the data of hosted plans is stored, was measured once: 630, 593, and 345 ms.
| Measurement origin | Zenifra API, São Paulo (median) | Washington, AWS us-east-1 (median) | Frankfurt, AWS eu-central-1 (1 sample) |
|---|---|---|---|
| São Paulo (Vivo) | 76 ms | 378 ms | 630 ms |
| Rio de Janeiro (Mundivox) | 77 ms | 335 ms | 593 ms |
| Fortaleza (IP Telecom) | 144 ms | 304 ms | 345 ms |
The effect grows with sequential calls, which is how many workflows work: receive the webhook, query an ERP, write to the database, call a messaging API. According to the same benchmark, on a connection that is already open each request costs about one round trip, 5 ms from São Paulo to São Paulo against 124 ms from São Paulo to Washington. Ten sequential calls add up to about 50 ms in the first case and 1,240 ms in the second, before any processing.
The database counts too. By default n8n saves executions to the database, so the application and PostgreSQL need to be in the same region; on Zenifra, both run in São Paulo. Read the numbers with the measurement’s limits in mind: one probe per city, an overnight window, and Frankfurt with a single sample. And the advantage only shows up if the service that sends the webhook also runs nearby, so check where the servers of each integration you use are located.
Is self-hosted n8n free to use at a company?
Yes, for internal use: n8n’s Sustainable Use License lets you use and modify the software free of charge for your own internal business purposes, for personal use, or for non-commercial use. You pay for the infrastructure, which on Zenifra comes to R$ 76.00 per month on the monthly contract with the template’s setup, not for a per-execution license.
The limit is offering n8n as a product. According to the license FAQ, making n8n available so your customers can connect their own accounts and build workflows requires a commercial agreement with n8n, while offering consulting and building workflows for clients is allowed without a separate agreement. Features such as SSO, multiple environments, and Git version control are on the Business plan, a paid license for self-hosted installs; if you are unsure about your case, n8n points to the license@n8n.io address.
Can I restrict the n8n editor by IP on Zenifra?
Yes, from the Premium Plus plan up, which offers an IP allowlist and blocklist for public projects; Basic, the template default, and Premium do not have this feature. But the rule applies to the whole project: once the allowlist has entries, only the listed IPs reach the application, and that includes webhook requests.
In practice, restricting by IP only works if you know the outbound IPs of every service that calls your webhooks and they fit within the list’s 10 entries. While access is restricted, the certificate for a custom domain is not issued either. For most installs, the better path is to keep the project public, create the owner account on first access, and require authentication on webhooks: the Webhook node supports Basic auth, Header auth, and JWT for any service that calls the URL.
How do I set n8n to Brasília time?
Add GENERIC_TIMEZONE=America/Sao_Paulo to the project’s environment variables, under Edit project; the template does not define this variable. Without it, n8n uses America/New_York, and a schedule for 8 a.m. fires at 8 a.m. New York time.
The Schedule node uses this timezone to decide when the workflow starts. n8n’s Docker install documentation also mentions TZ, which sets the system timezone and affects what scripts and commands such as date return; add it as well if any workflow runs scripts that depend on local time. Saving the variables restarts the instances. Unlike Zenifra’s Scheduled Jobs, which always read cron expressions in UTC, scheduling inside n8n follows the timezone you configure.
Can I run n8n on several instances or with auto-scaling on Zenifra?
Not with the template’s setup: it uses 1 instance, and running several copies of n8n side by side requires queue mode, which the template does not configure. For more capacity, move up a plan instead of turning on auto-scaling: Premium (1 vCPU and 1 GB) or Premium Plus (2 vCPU and 2 GB).
In queue mode, a main instance receives triggers and passes each execution ID to Redis, and workers pick executions off the queue and run them. The n8n documentation says queue mode with SQLite is not recommended, that the encryption key must be the same on the main instance and on every worker, and that filesystem binary data storage does not work in this mode, with S3 external storage as the alternative. This setup is not part of the template; the ready-made path is a single instance on a plan of the right size.
How do I use my own domain for n8n webhooks?
Add the domain under Edit, in the Custom domains section, create the DNS records the Console shows and, once the certificate is active, update the address in N8N_WEBHOOK_URL. From then on, the Webhook node shows URLs on your domain.
The Console shows the records table with the Type, Name, Create with value, and We found columns, refreshes the status every 30 seconds, and tells you when the HTTPS certificate is active. Each project accepts up to 20 custom domains, one new domain per change, and only public projects accept domains. External services store the full URL of each webhook, so update, in their dashboards, any webhooks registered with the old address.