MinIO — Browse the object storage without leaving the console

Checking whether a file made it into the platform's object storage becomes a folder walk inside DATTA itself, under the platform's permissions and audit trail. It used to require a command-line tool, credentials copied back and forth and an external console. The Sistema›MinIO screen brings together access configuration, the connectivity test and browsing of buckets and objects — including the Iceberg warehouse (datta-warehouse) that Apache Trino uses, plus the document and staging buckets.

The screen has three tabs:

Conexão tab

Fill in the endpoint, the access key and the secret key, the default bucket, the region and the path-style option. Once saved, the secret key shows up masked — it cannot be read back in clear text from the screen. Use Salvar configuração to store it and Testar conectividade to validate it right away.

The configuration is kept as a platform secret, under the name datta-minio-config.

Status tab

Shows the result of the connectivity check against the configured endpoint — the test that answers "is the storage reachable from the platform?" before you investigate anything else.

Buckets & Objetos tab

• Bucket list — cards with name and creation date; click to open.
• Folder navigation — the listing respects the folder hierarchy (the / delimiter), and the trail (bucket › folder › subfolder) takes you back to any level with one click.
• Bucket summary — object count and total size. The scan is limited to 50,000 objects; above that the total appears with a +.
• Object detail — clicking a file opens a side panel with key, size, Content-Type, modification date, ETag and the user metadata.
• Read-only — the browser does no upload, download or deletion. Content mutations belong to the platform's ingestion tools (Extract, DATTABI) and, in documented operational procedures, to the storage's own command-line client. Here you inspect safely, with no risk of erasing anything.

Worked example — checking that a Parquet reached the warehouse

1. Run your materialization flow in Extract and note the target dataset.
2. Open Sistema›MinIO › Buckets & Objetos and click the warehouse bucket (the same one shown as the default bucket in the Conexão tab).
3. Browse the dataset folder down to the .parquet files and check, in the detail panel, the size and the modification date — if the date matches the flow run, the materialization reached its destination.
4. Something odd? The trail lets you go up one level and compare with the other folders without losing context.

Where the credentials come from

Browsing resolves credentials in this order, never recording them in clear text in the logs:

1. The datta-minio-config secret, managed by the Conexão tab — used when the access key and secret key are filled in there.
2. The storage credentials provisioned at installation (MINIO_ACCESS_KEY, MINIO_SECRET_KEY, MINIO_ENDPOINT), available to every module of the platform — the same ones Trino uses for the warehouse.

If neither source has credentials, browsing returns a friendly error asking you to configure access in Sistema›MinIO › Conexão.

Permissions

A denial returns the message "Acesso negado." in Portuguese. Enforcement is always on the server; the interface only hides or disables what you cannot use. See the roles and permissions guide.

Listing cache

So as not to hit the storage on every click, reads are cached for a few seconds:

Automation

Everything the screen does — listing buckets, listing objects and folders, reading an object's detail, computing the bucket summary, reading/writing the access configuration and testing connectivity — is also available to verification automations; see the API reference.

Troubleshooting

See also

• Object storage console architecture
• SIQL storage lineage — consumes the same warehouse.
• API reference