# Dell ObjectScale (ECS) — atc-objectscale | Item | Value | |------|-------| | Hostname | `luna.local` / `atc-objectscale` | | IP | `10.0.20.111` | | SSH | `admin@10.0.20.111` (appliance; root via sudo keys) | | Management UI | https://10.0.20.111/ (port **443**) | | S3 API | http://10.0.20.111:**9020** | | Container | `ecs-storageos` (`emccorp/ecs-software`) | | Install config | `/opt/emc/ecs-install/deploy.yml` | ## What is `management_clients: 0.0.0.0/0`? In `deploy.yml`, this setting controls **which client IP addresses may access ECS management ports** (admin API, node management, not the S3 data path). | Value | Meaning | |-------|---------| | `0.0.0.0/0` | **Everyone on any network** — entire Internet can reach management ports if routed/firewall allows | | `10.0.20.0/24` | Only hosts in `10.0.20.x` subnet | | `10.0.21.45` | Only that single host | **Lab default:** installer often sets `0.0.0.0/0` for convenience (= no IP whitelist). **Production:** restrict to management subnet only, e.g.: ```yaml management_clients: - 10.0.10.0/24 # Proxmox / mgmt - 10.0.20.0/24 # storage VLAN - 10.0.21.0/24 # compute VLAN ``` Changing this requires editing `/opt/emc/ecs-install/deploy.yml` on the appliance and may need an ECS config apply — plan a maintenance window. ## S3 usage in this lab | Consumer | Bucket / path | Config | |----------|---------------|--------| | Trino Iceberg | via `iceberg.properties` | `hive.s3.endpoint=http://10.0.20.111:9020` | | Spark jobs | bucket `data` | `/opt/spark-jobs/*.py` on lake01 | | AWS CLI | `aws --endpoint-url http://10.0.20.111:9020` | keys in Trino catalog (redacted in git) | ## Ports (reference) | Port | Service | |------|---------| | 443 | ECS Management UI (HTTPS) | | 9020 | S3 API | | 4443 | ECS API (nginx) | | 9021–9025 | Data head services (internal) | ## Health check ```bash curl -sk -o /dev/null -w "%{http_code}\n" https://10.0.20.111/ # expect 200 curl -s -o /dev/null -w "%{http_code}\n" http://10.0.20.111:9020/ # expect 403 without auth (normal) ```