Edit on val.town ↗

Migrate blobs

Use this as a reference, not a required sequence. Adapt the approach and code to suit the app and the user’s needs.

Static assets that ship with the code can use the site's publish directory. For uploads, choose S3-compatible object storage or a Render persistent disk. Object storage is easier to share across web services and jobs; a disk belongs to one instance.

List the Val Town store and prefixes the app actually uses. Account-global storage may include unrelated apps. Use blob.list(prefix) to list keys and blob.get(key) to read each object’s bytes in the source val. Use the scoped/global import the app uses. Copy keys and bytes; preserve content types/metadata where present or derive them from app records/file types. Do not copy only download URLs.

Set the bucket and provider credentials as Render env vars. For a small file, @aws-sdk/client-s3 can upload it:

import { S3Client, PutObjectCommand, GetObjectCommand } from "@aws-sdk/client-s3";
import { readFile } from "node:fs/promises";

const s3 = new S3Client({
  region: process.env.S3_REGION,
  endpoint: process.env.S3_ENDPOINT,
  forcePathStyle: process.env.S3_FORCE_PATH_STYLE === "true",
  credentials: {
    accessKeyId: process.env.S3_ACCESS_KEY_ID,
    secretAccessKey: process.env.S3_SECRET_ACCESS_KEY,
  },
});
await s3.send(new PutObjectCommand({
  Bucket: process.env.S3_BUCKET, Key: "uploads/example.png",
  Body: await readFile("example.png"), ContentType: "image/png",
}));
const object = await s3.send(new GetObjectCommand({
  Bucket: process.env.S3_BUCKET, Key: "uploads/example.png",
}));
const downloaded = await object.Body.transformToByteArray();

Set the actual region and endpoint for the provider (auto is an R2 value, not a generic default). Use path-style addressing if the provider requires it. For large files, stream or use multipart upload rather than reading everything into memory. Replace Val Town's JSON helpers with explicit serialization/parsing.

For disk storage, write beneath its mount path and serve through the application. Other services cannot access that disk.

Keep private objects private; use authenticated routes or short-lived signed URLs. Update stored URLs and CORS where required. Copy in bounded batches with checkpoints and retries.

Compare counts, sizes, and content hashes from destination readback. Test read/write/delete and denied private access. Pause old writes, copy final changes and deletions, then cut over.

Last updated