(core) Prune snapshots outside the window in product features

Summary:
- Add a method `getSnapshotWindow` to `IInventory` and `DocSnapshotInventory`. It returns a `SnapshotWindow`, which represents a duration of time for which we keep backups for a particular document.
- `DocSnapshotPruner` calls this method and passes the window to `shouldKeepSnapshots` to determine which document versions have fallen outside the window and should be pruned.
- The implementation passed to `DocSnapshotInventory` uses a new method `getDocProduct` in `HomeDBManager` which directly returns the `Product` associated with a document, given only the document ID. Other methods in `HomeDBManager` require passing more information, especially about a user, but `DocSnapshotPruner` only knows about document IDs.

Test Plan: Added a test for `getDocProduct` and a test for `DocSnapshotPruner` where `getSnapshotWindow` is specified.

Reviewers: paulfitz

Reviewed By: paulfitz

Differential Revision: https://phab.getgrist.com/D3322
This commit is contained in:
Alex Hall 2022-03-15 12:45:20 +02:00
parent 21f1dfa56c
commit ec8460b772
5 changed files with 64 additions and 21 deletions

View File

@ -1,3 +1,8 @@
export interface SnapshotWindow {
count: number;
unit: 'month' | 'year';
}
// A product is essentially a list of flags and limits that we may enforce/support.
export interface Features {
vanityDomain?: boolean; // are user-selected domains allowed (unenforced) (default: true)
@ -35,10 +40,7 @@ export interface Features {
readOnlyDocs?: boolean; // if set, docs can only be read, not written.
snapshotWindow?: { // if set, controls how far back snapshots are kept.
count: number; // TODO: not honored at time of writing.
unit: 'month'|'year';
};
snapshotWindow?: SnapshotWindow; // if set, controls how far back snapshots are kept.
baseMaxRowsPerDocument?: number; // If set, establishes a default maximum on the
// number of rows (total) in a single document.

View File

@ -1,7 +1,8 @@
import {Features} from 'app/common/Features';
import {nativeValues} from 'app/gen-server/lib/values';
import * as assert from 'assert';
import {BaseEntity, Column, Connection, Entity, PrimaryGeneratedColumn} from 'typeorm';
import {BillingAccount} from 'app/gen-server/entity/BillingAccount';
import {BaseEntity, Column, Connection, Entity, OneToMany, PrimaryGeneratedColumn} from 'typeorm';
/**
* A summary of features used in 'starter' plans.
@ -149,6 +150,9 @@ export class Product extends BaseEntity {
@Column({type: nativeValues.jsonEntityType})
public features: Features;
@OneToMany(type => BillingAccount, account => account.product)
public accounts: BillingAccount[];
}
/**

View File

@ -2325,6 +2325,18 @@ export class HomeDBManager extends EventEmitter {
});
}
public async getDocProduct(docId: string): Promise<Product | undefined> {
return await this._connection.createQueryBuilder()
.select('product')
.from(Product, 'product')
.leftJoinAndSelect('product.accounts', 'account')
.leftJoinAndSelect('account.orgs', 'org')
.leftJoinAndSelect('org.workspaces', 'workspace')
.leftJoinAndSelect('workspace.docs', 'doc')
.where('doc.id = :docId', {docId})
.getOne();
}
/**
* Get the anonymous user, as a constructed object rather than a database lookup.
*/

View File

@ -1,7 +1,8 @@
import { ObjSnapshotWithMetadata } from 'app/common/DocSnapshot';
import { KeyedOps } from 'app/common/KeyedOps';
import { KeyedMutex } from 'app/common/KeyedMutex';
import { ExternalStorage } from 'app/server/lib/ExternalStorage';
import {ObjSnapshotWithMetadata} from 'app/common/DocSnapshot';
import {SnapshotWindow} from 'app/common/Features';
import {KeyedMutex} from 'app/common/KeyedMutex';
import {KeyedOps} from 'app/common/KeyedOps';
import {ExternalStorage} from 'app/server/lib/ExternalStorage';
import * as log from 'app/server/lib/log';
import * as fse from 'fs-extra';
import * as moment from 'moment-timezone';
@ -10,6 +11,7 @@ import * as moment from 'moment-timezone';
* A subset of the ExternalStorage interface, focusing on maintaining a list of versions.
*/
export interface IInventory {
getSnapshotWindow?: (key: string) => Promise<SnapshotWindow|undefined>;
versions(key: string): Promise<ObjSnapshotWithMetadata[]>;
remove(key: string, snapshotIds: string[]): Promise<void>;
}
@ -59,8 +61,9 @@ export class DocSnapshotPruner {
// Get all snapshots for a document, and whether they should be kept or pruned.
public async classify(key: string): Promise<Array<{snapshot: ObjSnapshotWithMetadata, keep: boolean}>> {
const snapshotWindow = await this._ext.getSnapshotWindow?.(key);
const versions = await this._ext.versions(key);
return shouldKeepSnapshots(versions).map((keep, index) => ({keep, snapshot: versions[index]}));
return shouldKeepSnapshots(versions, snapshotWindow).map((keep, index) => ({keep, snapshot: versions[index]}));
}
// Prune the specified document immediately. If no snapshotIds are provided, they
@ -107,8 +110,12 @@ export class DocSnapshotInventory implements IInventory {
* Expects to be given the store for documents, a store for metadata, and a method
* for naming cache files on the local filesystem. The stores should be consistent.
*/
constructor(private _doc: ExternalStorage, private _meta: ExternalStorage,
private _getFilename: (key: string) => Promise<string>) {}
constructor(
private _doc: ExternalStorage,
private _meta: ExternalStorage,
private _getFilename: (key: string) => Promise<string>,
public getSnapshotWindow: (key: string) => Promise<SnapshotWindow|undefined>,
) {}
/**
* Start keeping inventory for a new document.
@ -312,7 +319,7 @@ export class DocSnapshotInventory implements IInventory {
* - Anything with a label, for up to 32 days before the current version.
* Calculations done in UTC, Gregorian calendar, ISO weeks (week starts with Monday).
*/
export function shouldKeepSnapshots(snapshots: ObjSnapshotWithMetadata[]): boolean[] {
export function shouldKeepSnapshots(snapshots: ObjSnapshotWithMetadata[], snapshotWindow?: SnapshotWindow): boolean[] {
// Get current version
const current = snapshots[0];
if (!current) { return []; }
@ -334,15 +341,26 @@ export function shouldKeepSnapshots(snapshots: ObjSnapshotWithMetadata[]): boole
// For each snapshot starting with newest, check if it is worth saving by comparing
// it with the last saved snapshot based on hour, day, week, month, year
return snapshots.map((snapshot, index) => {
let keep = index < 5; // Keep 5 most recent versions
// Just to make extra sure we don't delete everything
if (index === 0) {
return true;
}
const date = moment.tz(snapshot.lastModified, tz);
// Limit snapshots to the given window corresponding to what the user has paid for
if (snapshotWindow && start.diff(date, snapshotWindow.unit, true) > snapshotWindow.count) {
return false;
}
let keep = index < 5; // Keep 5 most recent versions
for (const bucket of buckets) {
if (updateAndCheckRange(date, bucket)) { keep = true; }
}
// Preserve recent labelled snapshots in a naive and limited way. No doubt this will
// be elaborated on if we make this a user-facing feature.
if (snapshot.metadata?.label &&
start.diff(moment.tz(snapshot.lastModified, tz), 'days') < 32) { keep = true; }
start.diff(date, 'days') < 32) { keep = true; }
return keep;
});
}

View File

@ -162,12 +162,19 @@ export class HostedStorageManager implements IDocStorageManager {
this._latestMetaVersions,
options);
this._inventory = new DocSnapshotInventory(this._ext, this._extMeta,
async docId => {
const dir = this.getAssetPath(docId);
await fse.mkdirp(dir);
return path.join(dir, 'meta.json');
});
this._inventory = new DocSnapshotInventory(
this._ext,
this._extMeta,
async docId => {
const dir = this.getAssetPath(docId);
await fse.mkdirp(dir);
return path.join(dir, 'meta.json');
},
async docId => {
const product = await dbManager.getDocProduct(docId);
return product?.features.snapshotWindow;
},
);
// The pruner could use an inconsistent store without any real loss overall,
// but tests are easier if it is consistent.