---
title: "Object Storage"
description: "S3-compatible buckets and objects, used with standard S3 tools by changing the endpoint and keys."
---

> Documentation Index
> Fetch the complete documentation index at: https://docs.ezghcloud.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Object Storage

EZGH Object Storage stores objects in buckets and speaks the S3 API. AWS CLI, AWS SDKs, rclone and
other S3 tools work by changing the endpoint and keys.

| Setting | Value |
| --- | --- |
| Endpoint | `https://s3.us-west-1.ezghapps.com` |
| Virtual-hosted endpoint | `https://<bucket>.s3.us-west-1.ezghapps.com` |
| Region | `us-west-1` |
| Signature | SigV4 only. SigV2 answers `400 InvalidRequest` |

In the console, open **Object Storage**: **Buckets** holds your buckets and their objects, and
**Access keys** the keys S3 tools sign with.

- [Quickstart](/object-storage/quickstart) — Create a bucket and a key, then upload and download an object.
- [Access keys](/object-storage/access-keys) — Create and revoke keys, and the permissions they act with.
- [Public buckets and presigned URLs](/object-storage/public-buckets-and-presigned-urls) — Share objects with people who have no EZGH account.
- [Lifecycle rules](/object-storage/lifecycle-rules) — Delete objects by age or on a date, and abort unfinished uploads sooner.
- [Limits](/object-storage/limits) — Default limits, and what happens over them.

## Buckets

A bucket belongs to a project. Its name is unique across all customers:

- 3 to 63 lowercase letters, digits and hyphens, starting and ending with a letter or digit. No dots.
- `admin`, `swift`, `auth`, `health`, `ready` and `metrics` are unavailable.

Create buckets in the console, with [CreateBucket](/objects-api/CreateBucket/) in the Object
Storage API, or with S3's `CreateBucket`, which puts the bucket in the access key's project. A bucket must be empty to be deleted. Its name then stays reserved for your organization
for 30 days, and anyone else's `CreateBucket` for it answers `BucketAlreadyExists`.

## Supported S3 operations

| Area | Operations |
| --- | --- |
| Buckets | `ListBuckets`, `CreateBucket`, `DeleteBucket`, `HeadBucket`, `GetBucketLocation` |
| Objects | `PutObject`, `GetObject` (with `Range`), `HeadObject`, `DeleteObject`, `DeleteObjects`, `CopyObject`, `ListObjects`, `ListObjectsV2` |
| Multipart uploads | `CreateMultipartUpload`, `UploadPart`, `UploadPartCopy`, `CompleteMultipartUpload`, `AbortMultipartUpload`, `ListParts`, `ListMultipartUploads` |
| CORS | `GetBucketCors`, `PutBucketCors`, `DeleteBucketCors` |
| Versioning | `GetBucketVersioning`, `PutBucketVersioning`, `ListObjectVersions`, and `versionId` on `GetObject`, `HeadObject`, `DeleteObject`, `DeleteObjects` and copy sources. See [Versioning](/object-storage/versioning) |
| Lifecycle | `GetBucketLifecycleConfiguration`, `PutBucketLifecycleConfiguration`, `DeleteBucketLifecycle`. See [Lifecycle rules](/object-storage/lifecycle-rules) for the rules they take |

`CopyObject` and `UploadPartCopy` copy within your organization, and need download access to the
source bucket and upload access to the destination.

Everything else answers `NotImplemented`: bucket policies, ACLs other than `private`, MFA Delete,
object lock, replication, static website hosting, event notifications and S3 Select.
So does an `x-amz-` request header these operations don't use, such as `x-amz-tagging`.

- `x-amz-acl: private` is accepted and changes nothing. Any other ACL answers `NotImplemented`.
- `STANDARD` is the only storage class. Any other class answers `NotImplemented`.
- SigV2, in a header or a URL, answers `400 InvalidRequest`.

## Conditional writes

`PutObject` and `CompleteMultipartUpload` take these headers. A failed condition answers `412` and
leaves the object as it was.

| Header | Writes when |
| --- | --- |
| `If-None-Match: *` | The key doesn't exist |
| `If-Match: <ETag>` | The key's current ETag matches. A key that doesn't exist answers `404` |

Of concurrent `If-None-Match: *` writes of one key, exactly one succeeds, so lock files work. A
write that arrives while another conditional write of the same key is in progress answers
`409 ConditionalRequestConflict`: retry it. With
[boto3](/object-storage/boto3):

```python
from botocore.exceptions import ClientError

try:
    s3.put_object(Bucket="my-bucket", Key="locks/nightly", Body=b"worker-1", IfNoneMatch="*")
except ClientError as error:
    if error.response["ResponseMetadata"]["HTTPStatusCode"] not in (409, 412):
        raise
    print("Another worker holds the lock")
```

## Console

A bucket's page shows its overview, permissions, CORS rules, [versioning](/object-storage/versioning),
[lifecycle rules](/object-storage/lifecycle-rules), usage and objects. In **Objects** you
browse prefixes, view metadata, and download and delete objects and their versions. Upload with an S3 tool.

## CLI

`ezgh objects` manages buckets, their settings, lifecycle rules and access keys. Objects themselves go through S3
tools.

```sh
ezgh objects buckets create my-bucket
ezgh objects keys create --description laptop --print-credentials >> ~/.aws/config
aws --profile ezgh s3 ls
```

See the [command reference](/cli/reference/objects).

## Network

Instances reach Object Storage through the public endpoint, and that traffic counts as internet
egress.

## Trails

Trails records changes to buckets, their settings, lifecycle rules and access keys, including those made through the
S3 API. Object reads and writes aren't recorded. Secrets are never recorded.

Source: https://docs.ezghcloud.com/object-storage/index.mdx
