2020-10-30 16:53:23 +00:00
|
|
|
import {ObjMetadata, ObjSnapshot, ObjSnapshotWithMetadata} from 'app/common/DocSnapshot';
|
2020-07-21 13:20:51 +00:00
|
|
|
import * as log from 'app/server/lib/log';
|
|
|
|
import {createTmpDir} from 'app/server/lib/uploads';
|
|
|
|
import {delay} from 'bluebird';
|
|
|
|
import * as fse from 'fs-extra';
|
|
|
|
import * as path from 'path';
|
|
|
|
|
|
|
|
// A special token representing a deleted document, used in places where a
|
|
|
|
// checksum is expected otherwise.
|
|
|
|
export const DELETED_TOKEN = '*DELETED*';
|
|
|
|
|
|
|
|
/**
|
|
|
|
* An external store for the content of files. The store may be either consistent
|
|
|
|
* or eventually consistent. Specifically, the `exists`, `download`, and `versions`
|
|
|
|
* methods may return somewhat stale data from time to time.
|
|
|
|
*
|
|
|
|
* The store should be versioned; that is, uploads to a `key` should be assigned
|
|
|
|
* a `snapshotId`, and be accessible later with that `key`/`snapshotId` pair.
|
|
|
|
* When data is accessed by `snapshotId`, results should be immediately consistent.
|
|
|
|
*/
|
|
|
|
export interface ExternalStorage {
|
|
|
|
// Check if content exists in the store for a given key.
|
2020-10-30 16:53:23 +00:00
|
|
|
exists(key: string, snapshotId?: string): Promise<boolean>;
|
|
|
|
|
|
|
|
// Get side information for content, if content exists in the store.
|
|
|
|
head(key: string, snapshotId?: string): Promise<ObjSnapshotWithMetadata|null>;
|
2020-07-21 13:20:51 +00:00
|
|
|
|
|
|
|
// Upload content from file to the given key. Returns a snapshotId if store supports that.
|
2022-04-22 18:07:14 +00:00
|
|
|
upload(key: string, fname: string, metadata?: ObjMetadata): Promise<string|null|typeof Unchanged>;
|
2020-07-21 13:20:51 +00:00
|
|
|
|
|
|
|
// Download content from key to given file. Can download a specific version of the key
|
|
|
|
// if store supports that (should throw a fatal exception if not).
|
2020-11-10 03:28:30 +00:00
|
|
|
// Returns snapshotId of version downloaded.
|
|
|
|
download(key: string, fname: string, snapshotId?: string): Promise<string>;
|
2020-07-21 13:20:51 +00:00
|
|
|
|
|
|
|
// Remove content for this key from the store, if it exists. Can delete specific versions
|
|
|
|
// if specified. If no version specified, all versions are removed. If versions specified,
|
|
|
|
// newest should be given first.
|
|
|
|
remove(key: string, snapshotIds?: string[]): Promise<void>;
|
|
|
|
|
|
|
|
// List content versions that exist for the given key. More recent versions should
|
|
|
|
// come earlier in the result list.
|
|
|
|
versions(key: string): Promise<ObjSnapshot[]>;
|
|
|
|
|
|
|
|
// Render the given key as something url-like, for log messages (e.g. "s3://bucket/path")
|
|
|
|
url(key: string): string;
|
|
|
|
|
|
|
|
// Check if an exception thrown by a store method should be treated as fatal.
|
|
|
|
// Non-fatal exceptions are those that may result from eventual consistency, and
|
|
|
|
// where a retry could help -- specifically "not found" exceptions.
|
|
|
|
isFatalError(err: any): boolean;
|
|
|
|
|
|
|
|
// Close the storage object.
|
|
|
|
close(): Promise<void>;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Convenience wrapper to transform keys for an external store.
|
|
|
|
* E.g. this could convert "<docId>" to "v1/<docId>.grist"
|
|
|
|
*/
|
|
|
|
export class KeyMappedExternalStorage implements ExternalStorage {
|
|
|
|
constructor(private _ext: ExternalStorage,
|
|
|
|
private _map: (key: string) => string) {
|
|
|
|
}
|
|
|
|
|
2020-10-30 16:53:23 +00:00
|
|
|
public exists(key: string, snapshotId?: string): Promise<boolean> {
|
|
|
|
return this._ext.exists(this._map(key), snapshotId);
|
|
|
|
}
|
|
|
|
|
|
|
|
public head(key: string, snapshotId?: string) {
|
|
|
|
return this._ext.head(this._map(key), snapshotId);
|
2020-07-21 13:20:51 +00:00
|
|
|
}
|
|
|
|
|
2020-10-30 16:53:23 +00:00
|
|
|
public upload(key: string, fname: string, metadata?: ObjMetadata) {
|
|
|
|
return this._ext.upload(this._map(key), fname, metadata);
|
2020-07-21 13:20:51 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
public download(key: string, fname: string, snapshotId?: string) {
|
|
|
|
return this._ext.download(this._map(key), fname, snapshotId);
|
|
|
|
}
|
|
|
|
|
|
|
|
public remove(key: string, snapshotIds?: string[]): Promise<void> {
|
|
|
|
return this._ext.remove(this._map(key), snapshotIds);
|
|
|
|
}
|
|
|
|
|
|
|
|
public versions(key: string) {
|
|
|
|
return this._ext.versions(this._map(key));
|
|
|
|
}
|
|
|
|
|
|
|
|
public url(key: string) {
|
|
|
|
return this._ext.url(this._map(key));
|
|
|
|
}
|
|
|
|
|
|
|
|
public isFatalError(err: any) {
|
|
|
|
return this._ext.isFatalError(err);
|
|
|
|
}
|
|
|
|
|
|
|
|
public async close() {
|
|
|
|
// nothing to do
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* A wrapper for an external store that uses checksums and retries
|
|
|
|
* to compensate for eventual consistency. With this wrapper, the
|
|
|
|
* store either returns consistent results or fails with an error.
|
|
|
|
*
|
|
|
|
* This wrapper works by tracking what is in the external store,
|
|
|
|
* using content hashes and ids placed in consistent stores. These
|
|
|
|
* consistent stores are:
|
|
|
|
*
|
|
|
|
* - sharedHash: a key/value store containing expected checksums
|
|
|
|
* of content in the external store. In our setup, this is
|
|
|
|
* implemented using Redis. Populated on upload and checked on
|
|
|
|
* download.
|
|
|
|
* - localHash: a key/value store containing checksums of uploaded
|
|
|
|
* content. In our setup, this is implemented on the worker's
|
|
|
|
* disk. This is used to skip unnecessary uploads. Populated
|
|
|
|
* on download and checked on upload.
|
|
|
|
* - latestVersion: a key/value store containing snapshotIds of
|
|
|
|
* uploads. In our setup, this is implemented in the worker's
|
|
|
|
* memory. Only affects the consistency of the `versions` method.
|
|
|
|
* Populated on upload and checked on `versions` calls.
|
|
|
|
* TODO: move to Redis if consistency of `versions` during worker
|
|
|
|
* transitions becomes important.
|
|
|
|
*
|
|
|
|
* It is not important for all this side information to persist very
|
|
|
|
* long, just long enough to give the store time to become
|
|
|
|
* consistent.
|
|
|
|
*
|
|
|
|
* Keys presented to this class should be file-system safe.
|
|
|
|
*/
|
|
|
|
export class ChecksummedExternalStorage implements ExternalStorage {
|
|
|
|
private _closed: boolean = false;
|
|
|
|
|
2021-04-26 21:54:09 +00:00
|
|
|
constructor(public readonly label: string, private _ext: ExternalStorage, private _options: {
|
2020-07-21 13:20:51 +00:00
|
|
|
maxRetries: number, // how many time to retry inconsistent downloads
|
|
|
|
initialDelayMs: number, // how long to wait before retrying
|
|
|
|
localHash: PropStorage, // key/value store for hashes of downloaded content
|
|
|
|
sharedHash: PropStorage, // key/value store for hashes of external content
|
|
|
|
latestVersion: PropStorage, // key/value store for snapshotIds of uploads
|
|
|
|
computeFileHash: (fname: string) => Promise<string>, // compute hash for file
|
|
|
|
}) {
|
|
|
|
}
|
|
|
|
|
2020-10-30 16:53:23 +00:00
|
|
|
public async exists(key: string, snapshotId?: string): Promise<boolean> {
|
|
|
|
return this._retryWithExistenceCheck('exists', key, snapshotId,
|
|
|
|
this._ext.exists.bind(this._ext));
|
2020-07-21 13:20:51 +00:00
|
|
|
}
|
|
|
|
|
2020-10-30 16:53:23 +00:00
|
|
|
public async head(key: string, snapshotId?: string) {
|
|
|
|
return this._retryWithExistenceCheck('head', key, snapshotId,
|
|
|
|
this._ext.head.bind(this._ext));
|
|
|
|
}
|
|
|
|
|
|
|
|
public async upload(key: string, fname: string, metadata?: ObjMetadata) {
|
2020-07-21 13:20:51 +00:00
|
|
|
try {
|
|
|
|
const checksum = await this._options.computeFileHash(fname);
|
|
|
|
const prevChecksum = await this._options.localHash.load(key);
|
2020-10-30 16:53:23 +00:00
|
|
|
if (prevChecksum && prevChecksum === checksum && !metadata?.label) {
|
2020-07-21 13:20:51 +00:00
|
|
|
// nothing to do, checksums match
|
2020-11-10 03:28:30 +00:00
|
|
|
const snapshotId = await this._options.latestVersion.load(key);
|
|
|
|
log.info("ext %s upload: %s unchanged, not sending (checksum %s, version %s)", this.label, key,
|
|
|
|
checksum, snapshotId);
|
2022-04-22 18:07:14 +00:00
|
|
|
return Unchanged;
|
2020-07-21 13:20:51 +00:00
|
|
|
}
|
2020-10-30 16:53:23 +00:00
|
|
|
const snapshotId = await this._ext.upload(key, fname, metadata);
|
2020-11-10 03:28:30 +00:00
|
|
|
log.info("ext %s upload: %s checksum %s version %s", this.label, this._ext.url(key), checksum, snapshotId);
|
2022-04-22 18:07:14 +00:00
|
|
|
if (typeof snapshotId === "string") { await this._options.latestVersion.save(key, snapshotId); }
|
2020-07-21 13:20:51 +00:00
|
|
|
await this._options.localHash.save(key, checksum);
|
|
|
|
await this._options.sharedHash.save(key, checksum);
|
|
|
|
return snapshotId;
|
|
|
|
} catch (err) {
|
2020-11-10 03:28:30 +00:00
|
|
|
log.error("ext %s upload: %s failure to send, error %s", this.label, key, err.message);
|
2020-07-21 13:20:51 +00:00
|
|
|
throw err;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
public async remove(key: string, snapshotIds?: string[]) {
|
|
|
|
try {
|
|
|
|
// Removing most recent version by id is not something we should be doing, and
|
|
|
|
// if we want to do it it would need to be done carefully - so just forbid it.
|
|
|
|
if (snapshotIds && snapshotIds.includes(await this._options.latestVersion.load(key) || '')) {
|
|
|
|
throw new Error('cannot remove most recent version of a document by id');
|
|
|
|
}
|
|
|
|
await this._ext.remove(key, snapshotIds);
|
2020-11-10 03:28:30 +00:00
|
|
|
log.info("ext %s remove: %s version %s", this.label, this._ext.url(key), snapshotIds || 'ALL');
|
2020-07-21 13:20:51 +00:00
|
|
|
if (!snapshotIds) {
|
|
|
|
await this._options.latestVersion.save(key, DELETED_TOKEN);
|
|
|
|
await this._options.sharedHash.save(key, DELETED_TOKEN);
|
2021-05-23 17:43:11 +00:00
|
|
|
} else {
|
|
|
|
for (const snapshotId of snapshotIds) {
|
|
|
|
// Removing snapshots breaks their partial immutability, so we mark them
|
|
|
|
// as deleted in redis so that we don't get stale info from S3 if we check
|
|
|
|
// for their existence. Nothing currently depends on this in practice.
|
|
|
|
await this._options.sharedHash.save(this._keyWithSnapshot(key, snapshotId), DELETED_TOKEN);
|
|
|
|
}
|
2020-07-21 13:20:51 +00:00
|
|
|
}
|
|
|
|
} catch (err) {
|
2020-11-10 03:28:30 +00:00
|
|
|
log.error("ext %s delete: %s failure to remove, error %s", this.label, key, err.message);
|
2020-07-21 13:20:51 +00:00
|
|
|
throw err;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
public download(key: string, fname: string, snapshotId?: string) {
|
|
|
|
return this.downloadTo(key, key, fname, snapshotId);
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* We may want to download material from one key and henceforth treat it as another
|
2022-02-19 09:46:49 +00:00
|
|
|
* key (specifically for forking a document). Since this class cross-references the
|
2020-07-21 13:20:51 +00:00
|
|
|
* key in the external store with other consistent stores, it needs to know we are
|
|
|
|
* doing that. So we add a downloadTo variant that takes before and after keys.
|
|
|
|
*/
|
|
|
|
public async downloadTo(fromKey: string, toKey: string, fname: string, snapshotId?: string) {
|
2020-11-10 03:28:30 +00:00
|
|
|
return this._retry('download', async () => {
|
2020-07-21 13:20:51 +00:00
|
|
|
const {tmpDir, cleanupCallback} = await createTmpDir({});
|
2020-10-30 16:53:23 +00:00
|
|
|
const tmpPath = path.join(tmpDir, `${toKey}-tmp`); // NOTE: assumes key is file-system safe.
|
2020-07-21 13:20:51 +00:00
|
|
|
try {
|
2020-11-10 03:28:30 +00:00
|
|
|
const downloadedSnapshotId = await this._ext.download(fromKey, tmpPath, snapshotId);
|
2020-07-21 13:20:51 +00:00
|
|
|
|
|
|
|
const checksum = await this._options.computeFileHash(tmpPath);
|
|
|
|
|
|
|
|
// Check for consistency if mutable data fetched.
|
|
|
|
if (!snapshotId) {
|
|
|
|
const expectedChecksum = await this._options.sharedHash.load(fromKey);
|
|
|
|
// Let null docMD5s pass. Otherwise we get stuck if redis is cleared.
|
|
|
|
// Otherwise, make sure what we've got matches what we expect to get.
|
|
|
|
// S3 is eventually consistent - if you overwrite an object in it, and then read from it,
|
|
|
|
// you may get an old version for some time.
|
|
|
|
// If a snapshotId was specified, we can skip this check.
|
|
|
|
if (expectedChecksum && expectedChecksum !== checksum) {
|
2020-11-10 03:28:30 +00:00
|
|
|
log.error("ext %s download: data for %s has wrong checksum: %s (expected %s)",
|
|
|
|
this.label, fromKey, checksum, expectedChecksum);
|
2020-07-21 13:20:51 +00:00
|
|
|
return undefined;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// If successful, rename the temporary file to its proper name. The destination should NOT
|
|
|
|
// exist in this case, and this should fail if it does.
|
|
|
|
await fse.move(tmpPath, fname, {overwrite: false});
|
2020-11-10 03:28:30 +00:00
|
|
|
if (fromKey === toKey) {
|
2021-12-01 17:01:02 +00:00
|
|
|
// Save last S3 snapshot id observed for this key.
|
2020-11-10 03:28:30 +00:00
|
|
|
await this._options.latestVersion.save(toKey, downloadedSnapshotId);
|
2021-12-01 17:01:02 +00:00
|
|
|
// Save last S3 hash observed for this key (so if we have a version with that hash
|
|
|
|
// locally we can skip pushing it back needlessly later).
|
|
|
|
await this._options.localHash.save(toKey, checksum);
|
2020-11-10 03:28:30 +00:00
|
|
|
}
|
2020-07-21 13:20:51 +00:00
|
|
|
|
2020-11-10 03:28:30 +00:00
|
|
|
log.info("ext %s download: %s%s%s with checksum %s and version %s", this.label, fromKey,
|
2020-07-21 13:20:51 +00:00
|
|
|
snapshotId ? ` [VersionId ${snapshotId}]` : '',
|
|
|
|
fromKey !== toKey ? ` as ${toKey}` : '',
|
2020-11-10 03:28:30 +00:00
|
|
|
checksum, downloadedSnapshotId);
|
2020-07-21 13:20:51 +00:00
|
|
|
|
2020-11-10 03:28:30 +00:00
|
|
|
return downloadedSnapshotId;
|
2020-07-21 13:20:51 +00:00
|
|
|
} catch (err) {
|
2020-11-10 03:28:30 +00:00
|
|
|
log.error("ext %s download: failed to fetch data (%s): %s", this.label, fromKey, err.message);
|
2020-07-21 13:20:51 +00:00
|
|
|
throw err;
|
|
|
|
} finally {
|
|
|
|
await cleanupCallback();
|
|
|
|
}
|
|
|
|
});
|
|
|
|
}
|
|
|
|
|
|
|
|
public async versions(key: string) {
|
|
|
|
return this._retry('versions', async () => {
|
|
|
|
const snapshotId = await this._options.latestVersion.load(key);
|
|
|
|
if (snapshotId === DELETED_TOKEN) { return []; }
|
|
|
|
const result = await this._ext.versions(key);
|
|
|
|
if (snapshotId && (result.length === 0 || result[0].snapshotId !== snapshotId)) {
|
|
|
|
// Result is not consistent yet.
|
|
|
|
return undefined;
|
|
|
|
}
|
|
|
|
return result;
|
|
|
|
});
|
|
|
|
}
|
|
|
|
|
|
|
|
public url(key: string): string {
|
|
|
|
return this._ext.url(key);
|
|
|
|
}
|
|
|
|
|
|
|
|
public isFatalError(err: any): boolean {
|
|
|
|
return this._ext.isFatalError(err);
|
|
|
|
}
|
|
|
|
|
|
|
|
public async close() {
|
|
|
|
this._closed = true;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Call an operation until it returns a value other than undefined.
|
|
|
|
*
|
|
|
|
* While the operation returns undefined, it will be retried for some period.
|
|
|
|
* This period is chosen to be long enough for S3 to become consistent.
|
|
|
|
*
|
|
|
|
* If the operation throws an error, and that error is not fatal (as determined
|
|
|
|
* by `isFatalError`, then it will also be retried. Fatal errors are thrown
|
|
|
|
* immediately.
|
|
|
|
*
|
|
|
|
* Once the operation returns a result, we pass that along. If it fails to
|
|
|
|
* return a result after all the allowed retries, a special exception is thrown.
|
|
|
|
*/
|
|
|
|
private async _retry<T>(name: string, operation: () => Promise<T|undefined>): Promise<T> {
|
|
|
|
let backoffCount = 1;
|
|
|
|
let backoffFactor = this._options.initialDelayMs;
|
|
|
|
const problems = new Array<[number, string|Error]>();
|
|
|
|
const start = Date.now();
|
|
|
|
while (backoffCount <= this._options.maxRetries) {
|
|
|
|
try {
|
|
|
|
const result = await operation();
|
|
|
|
if (result !== undefined) { return result; }
|
|
|
|
problems.push([Date.now() - start, 'not ready']);
|
|
|
|
} catch (err) {
|
|
|
|
if (this._ext.isFatalError(err)) {
|
|
|
|
throw err;
|
|
|
|
}
|
|
|
|
problems.push([Date.now() - start, err]);
|
|
|
|
}
|
|
|
|
// Wait some time before attempting to reload from s3. The longer we wait, the greater
|
|
|
|
// the odds of success. In practice, a second should be more than enough almost always.
|
|
|
|
await delay(Math.round(backoffFactor));
|
|
|
|
if (this._closed) { throw new Error('storage closed'); }
|
|
|
|
backoffCount++;
|
|
|
|
backoffFactor *= 1.7;
|
|
|
|
}
|
|
|
|
log.error(`operation failed to become consistent: ${name} - ${problems}`);
|
|
|
|
throw new Error(`operation failed to become consistent: ${name} - ${problems}`);
|
|
|
|
}
|
2020-10-30 16:53:23 +00:00
|
|
|
|
|
|
|
/**
|
|
|
|
* Retry an operation which will fail if content does not exist, until it is consistent
|
|
|
|
* with our expectation of the content's existence.
|
|
|
|
*/
|
|
|
|
private async _retryWithExistenceCheck<T>(label: string, key: string, snapshotId: string|undefined,
|
|
|
|
op: (key: string, snapshotId?: string) => Promise<T>): Promise<T> {
|
|
|
|
return this._retry(label, async () => {
|
|
|
|
const hash = await this._options.sharedHash.load(this._keyWithSnapshot(key, snapshotId));
|
|
|
|
const expected = hash !== null && hash !== DELETED_TOKEN;
|
|
|
|
const reported = await op(key, snapshotId);
|
|
|
|
// If we expect an object but store doesn't seem to have it, retry.
|
|
|
|
if (expected && !reported) { return undefined; }
|
|
|
|
// If store says there is an object but that is not what we expected (if we
|
|
|
|
// expected anything), retry.
|
|
|
|
if (hash && !expected && reported) { return undefined; }
|
|
|
|
// If expectations are matched, or we don't have expectations, return.
|
|
|
|
return reported;
|
|
|
|
});
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Generate a key to use with Redis for a document. Add in snapshot information
|
|
|
|
* if that is present (snapshots are immutable, except that they can be deleted,
|
|
|
|
* so we only set checksums for them in Redis when they are deleted).
|
|
|
|
*/
|
|
|
|
private _keyWithSnapshot(key: string, snapshotId?: string|null) {
|
|
|
|
return snapshotId ? `${key}--${snapshotId}` : key;
|
|
|
|
}
|
2020-07-21 13:20:51 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Small interface for storing hashes and ids.
|
|
|
|
*/
|
|
|
|
export interface PropStorage {
|
|
|
|
save(key: string, val: string): Promise<void>;
|
|
|
|
load(key: string): Promise<string|null>;
|
|
|
|
}
|
2022-04-22 18:07:14 +00:00
|
|
|
|
|
|
|
export const Unchanged = Symbol('Unchanged');
|
(core) add machinery for self-managed flavor of Grist
Summary:
Currently, we have two ways that we deliver Grist. One is grist-core,
which has simple defaults and is relatively easy for third parties to
deploy. The second is our internal build for our SaaS, which is the
opposite. For self-managed Grist, a planned paid on-premise version
of Grist, I adopt the following approach:
* Use the `grist-core` build mechanism, extending it to accept an
overlay of extra code if present.
* Extra code is supplied in a self-contained `ext` directory, with
an `ext/app` directory that is of same structure as core `app`
and `stubs/app`.
* The `ext` directory also contains information about extra
node dependencies needed beyond that of `grist-core`.
* The `ext` directory is contained within our monorepo rather than
`grist-core` since it may contain material not under the Apache
license.
Docker builds are achieved in our monorepo by using the `--build-context`
functionality to add in `ext` during the regular `grist-core` build:
```
docker buildx build --load -t gristlabs/grist-ee --build-context=ext=../ext .
```
Incremental builds in our monorepo are achieved with the `build_core.sh` helper,
like:
```
buildtools/build_core.sh /tmp/self-managed
cd /tmp/self-managed
yarn start
```
The initial `ext` directory contains material for snapshotting to S3.
If you build the docker image as above, and have S3 access, you can
do something like:
```
docker run -p 8484:8484 --env GRIST_SESSION_SECRET=a-secret \
--env GRIST_DOCS_S3_BUCKET=grist-docs-test \
--env GRIST_DOCS_S3_PREFIX=self-managed \
-v $HOME/.aws:/root/.aws -it gristlabs/grist-ee
```
This will start a version of Grist that is like `grist-core` but with
S3 snapshots enabled. To release this code to `grist-core`, it would
just need to move from `ext/app` to `app` within core.
I tried a lot of ways of organizing self-managed Grist, and this was
what made me happiest. There are a lot of trade-offs, but here is what
I was looking for:
* Only OSS-code in grist-core. Adding mixed-license material there
feels unfair to people already working with the repo. That said,
a possible future is to move away from our private monorepo to
a public mixed-licence repo, which could have the same relationship
with grist-core as the monorepo has.
* Minimal differences between self-managed builds and one of our
existing builds, ideally hewing as close to grist-core as possible
for ease of documentation, debugging, and maintenance.
* Ideally, docker builds without copying files around (the new
`--build-context` functionality made that possible).
* Compatibility with monorepo build.
Expressing dependencies of the extra code in `ext` proved tricky to
do in a clean way. Yarn/npm fought me every step of the way - everything
related to optional dependencies was unsatisfactory in some respect.
Yarn2 is flexible but smells like it might be overreach. In the end,
organizing to install non-core dependencies one directory up from the
main build was a good simple trick that saved my bacon.
This diff gets us to the point of building `grist-ee` images conveniently,
but there isn't a public repo people can go look at to see its source. This
could be generated by taking `grist-core`, adding the `ext` directory
to it, and pushing to a distinct repository. I'm not in a hurry to do that,
since a PR to that repo would be hard to sync with our monorepo and
`grist-core`. Also, we don't have any licensing text ready for the `ext`
directory. So leaving that for future work.
Test Plan: manual
Reviewers: georgegevoian, alexmojaki
Reviewed By: georgegevoian, alexmojaki
Differential Revision: https://phab.getgrist.com/D3415
2022-05-12 15:24:48 +00:00
|
|
|
|
|
|
|
export interface ExternalStorageSettings {
|
|
|
|
purpose: 'doc' | 'meta';
|
|
|
|
basePrefix?: string;
|
|
|
|
extraPrefix?: string;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* The storage mapping we use for our SaaS. A reasonable default, but relies
|
|
|
|
* on appropriate lifecycle rules being set up in the bucket.
|
|
|
|
*/
|
|
|
|
export function getExternalStorageKeyMap(settings: ExternalStorageSettings): (docId: string) => string {
|
|
|
|
const {basePrefix, extraPrefix, purpose} = settings;
|
|
|
|
let fullPrefix = basePrefix + (basePrefix?.endsWith('/') ? '' : '/');
|
|
|
|
if (extraPrefix) {
|
|
|
|
fullPrefix += extraPrefix + (extraPrefix.endsWith('/') ? '' : '/');
|
|
|
|
}
|
|
|
|
|
|
|
|
// Set up how we name files/objects externally.
|
|
|
|
let fileNaming: (docId: string) => string;
|
|
|
|
if (purpose === 'doc') {
|
|
|
|
fileNaming = docId => `${docId}.grist`;
|
|
|
|
} else if (purpose === 'meta') {
|
|
|
|
// Put this in separate prefix so a lifecycle rule can prune old versions of the file.
|
|
|
|
// Alternatively, could go in separate bucket.
|
|
|
|
fileNaming = docId => `assets/unversioned/${docId}/meta.json`;
|
|
|
|
} else {
|
|
|
|
throw new Error('create.ExternalStorage: unrecognized purpose');
|
|
|
|
}
|
|
|
|
return docId => (fullPrefix + fileNaming(docId));
|
|
|
|
}
|
|
|
|
|
|
|
|
export function wrapWithKeyMappedStorage(rawStorage: ExternalStorage, settings: ExternalStorageSettings) {
|
|
|
|
return new KeyMappedExternalStorage(rawStorage, getExternalStorageKeyMap(settings));
|
|
|
|
}
|