> ## Documentation Index > Fetch the complete documentation index at: https://docs.blaxel.ai/llms.txt > Use this file to discover all available pages before exploring further. # Export the filesystem changes to a presigned URL > Archives everything the sandbox changed on top of its base image and streams it, uncompressed, to a presigned S3 PUT URL. The memory of the sandbox is not archived. The sandbox is quiesced first: the process list is saved (unless saveProcesses is false), every process is stopped, and the API then refuses the calls that would write to the filesystem. The freeze is not lifted afterwards, since an exported sandbox is meant to be restored elsewhere; call POST /archive/resume to lift it. Use dryRun to get the archive's content and exact size without stopping anything and without uploading. Set async to start the export and answer immediately, which is what archiving a large filesystem needs: the export then reports itself through GET /archive/status. ## OpenAPI ````yaml https://raw.githubusercontent.com/blaxel-ai/sandbox/refs/heads/main/sandbox-api/docs/openapi.yml post /archive/export openapi: 3.0.0 info: contact: {} description: API for manipulating filesystem, processes and network. title: Sandbox API version: 0.0.1 servers: - url: https://sbx-{sandbox_id}-{workspace_id}.{region}.bl.run variables: sandbox_id: default: unknown workspace_id: default: unknown region: default: unknown security: - BearerAuth: [] paths: /archive/export: post: tags: - archive summary: Export the filesystem changes to a presigned URL description: >- Archives everything the sandbox changed on top of its base image and streams it, uncompressed, to a presigned S3 PUT URL. The memory of the sandbox is not archived. The sandbox is quiesced first: the process list is saved (unless saveProcesses is false), every process is stopped, and the API then refuses the calls that would write to the filesystem. The freeze is not lifted afterwards, since an exported sandbox is meant to be restored elsewhere; call POST /archive/resume to lift it. Use dryRun to get the archive's content and exact size without stopping anything and without uploading. Set async to start the export and answer immediately, which is what archiving a large filesystem needs: the export then reports itself through GET /archive/status. requestBody: content: application/json: schema: $ref: '#/components/schemas/ExportOptions' description: Export options required: true responses: '200': description: Export result content: application/json: schema: $ref: '#/components/schemas/ExportResult' '202': description: The export was started and runs in the background content: application/json: schema: $ref: '#/components/schemas/ExportProgress' '400': description: Invalid request content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' '409': description: An export is already in progress content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' '500': description: Export failed content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' components: schemas: ExportOptions: properties: async: description: >- Async starts the export and answers immediately, leaving it to run: an archive of a large filesystem takes longer than a request may be held open. Its progress is reported by /archive/status. example: false type: boolean dryRun: description: |- DryRun reports what would be archived, and its exact size, without stopping anything and without uploading. example: false type: boolean excludes: description: Excludes are added to the paths excluded by default. items: type: string type: array headers: additionalProperties: type: string description: >- Headers are sent with the upload request as given. A presigned URL only accepts the headers it was signed for, so these have to match what the caller signed: sending one that was not signed, or signing one that is not sent, is rejected as a signature mismatch. Typical use is a storage class, x-amz-storage-class: GLACIER_IR. type: object imageDevice: description: >- ImageDevice is the device holding the pristine image, /dev/vda by default. A generation that attaches the image elsewhere names its device here. example: /dev/vda type: string imageMountPoint: description: >- ImageMountPoint is a directory where the pristine image is already mounted. When set the image device is neither mounted nor unmounted. example: /mnt/lower type: string multipart: allOf: - $ref: '#/components/schemas/MultipartUpload' description: >- Multipart uploads the archive part by part instead, which is how an archive larger than the 5 GB a single PUT accepts is stored. It takes precedence over URL. saveProcesses: description: |- SaveProcesses stores the process list in the archive so restore can relaunch the workload. Defaults to true; set it to false to archive storage only. example: true type: boolean stopTimeoutSeconds: description: StopTimeoutSeconds bounds the graceful stop of each process. example: 30 type: integer url: description: >- URL is a presigned S3 PUT URL the archive is streamed to. Empty is only valid with DryRun, or with Multipart. example: https://bucket.s3.amazonaws.com/key?... type: string type: object ExportResult: properties: changes: description: |- Changes lists every path in the archive. Only filled for a dry run, where it is the point of the call. items: $ref: '#/components/schemas/archive.Change' type: array duration: example: 4.2s type: string manifest: $ref: '#/components/schemas/ArchiveManifest' size: description: |- Size is the exact number of bytes uploaded, known before the upload starts since the archive is not compressed. example: 3074211 type: integer stoppedProcesses: description: StoppedProcesses are the processes stopped to freeze the filesystem. items: type: string type: array uploaded: description: Uploaded is false for a dry run. example: true type: boolean required: - manifest type: object ExportProgress: properties: error: description: Error is why the export failed, without the presigned URL it used. type: string finishedAt: type: string size: description: >- Size is the archive's exact size, known once the filesystem is scanned. example: 3074211 type: integer startedAt: type: string state: allOf: - $ref: '#/components/schemas/archive.ExportState' example: running uploaded: description: Uploaded reports whether the storage holds the archive. example: false type: boolean type: object ErrorResponse: properties: error: example: Error message type: string required: - error type: object MultipartUpload: properties: abortUrl: description: >- AbortURL is a presigned DELETE URL that discards the parts already uploaded. Without it a failed export leaves them on the storage until a lifecycle rule removes them. type: string completeUrl: description: CompleteURL is a presigned POST URL that assembles the parts. type: string partSize: description: PartSize is the number of bytes sent to every part but the last. example: 536870912 type: integer partUrls: description: >- PartURLs are presigned PUT URLs, one per part, in order. Extra ones are left unused. items: type: string type: array type: object archive.Change: properties: kind: allOf: - $ref: '#/components/schemas/archive.ChangeKind' example: added path: description: Path is relative to the root, without a leading slash. example: usr/bin/curl type: string size: description: >- Size is the file content size in bytes, 0 for anything but a regular file. example: 256216 type: integer required: - kind - path type: object ArchiveManifest: properties: added: description: Added and Modified count the archive's payload members. example: 11 type: integer apiVersion: description: APIVersion is the sandbox-api build that produced the archive. example: v0.1.0 type: string createdAt: type: string deleted: description: >- Deleted are paths the image has and the sandbox deleted. Tar cannot carry a deletion, so import applies these from the manifest. items: type: string type: array excludes: description: Excludes are the paths left out of the comparison. items: type: string type: array imageDevice: description: >- ImageDevice is the device the pristine image was read from, for the record: it says which generation exported the archive (/dev/vda on mk3.1). example: /dev/vda type: string modified: example: 3 type: integer payloadBytes: description: PayloadBytes is the total content size of the payload members. example: 3073449 type: integer processes: description: Processes tells whether ProcessesName is present. example: true type: boolean root: description: Root is the directory the paths are relative to. example: / type: string version: example: 1 type: integer required: - createdAt - root - version type: object archive.ExportState: enum: - running - succeeded - failed type: string x-enum-varnames: - ExportRunning - ExportSucceeded - ExportFailed archive.ChangeKind: enum: - added - modified - deleted type: string x-enum-varnames: - ChangeAdded - ChangeModified - ChangeDeleted securitySchemes: BearerAuth: type: http scheme: bearer bearerFormat: JWT ````