batch processing w/ exponential backoff (#79)
Signed-off-by: Brian DeHamer <bdehamer@github.com>
This commit is contained in:
15
README.md
15
README.md
@@ -65,7 +65,7 @@ See [action.yml](action.yml)
|
||||
with:
|
||||
# Path to the artifact serving as the subject of the attestation. Must
|
||||
# specify exactly one of "subject-path" or "subject-digest". May contain
|
||||
# a glob pattern or list of paths (total subject count cannot exceed 64).
|
||||
# a glob pattern or list of paths (total subject count cannot exceed 2500).
|
||||
subject-path:
|
||||
|
||||
# SHA256 digest of the subject for the attestation. Must be in the form
|
||||
@@ -115,6 +115,15 @@ If multiple subjects are being attested at the same time, each attestation will
|
||||
be written to the output file on a separate line (using the [JSON Lines][7]
|
||||
format).
|
||||
|
||||
## Attestation Limits
|
||||
|
||||
### Subject Limits
|
||||
|
||||
No more than 2500 subjects can be attested at the same time. Subjects will be
|
||||
processed in batches 50. After the initial group of 50, each subsequent batch
|
||||
will incur an exponentially increasing amount of delay (capped at 1 minute of
|
||||
delay per batch) to avoid overwhelming the attestation API.
|
||||
|
||||
## Examples
|
||||
|
||||
### Identify Subject by Path
|
||||
@@ -175,8 +184,8 @@ fully-qualified image name (e.g. "ghcr.io/user/app" or
|
||||
"acme.azurecr.io/user/app"). Do NOT include a tag as part of the image name --
|
||||
the specific image being attested is identified by the supplied digest.
|
||||
|
||||
> **NOTE**: When pushing to Docker Hub, please use "docker.io" as the
|
||||
> registry portion of the image name.
|
||||
> **NOTE**: When pushing to Docker Hub, please use "docker.io" as the registry
|
||||
> portion of the image name.
|
||||
|
||||
```yaml
|
||||
name: build-attested-image
|
||||
|
||||
@@ -46,8 +46,7 @@ const defaultInputs: main.RunInputs = {
|
||||
pushToRegistry: false,
|
||||
githubToken: '',
|
||||
privateSigning: false,
|
||||
batchSize: 50,
|
||||
batchDelay: 5000
|
||||
batchSize: 50
|
||||
}
|
||||
|
||||
describe('action', () => {
|
||||
@@ -355,8 +354,7 @@ describe('action', () => {
|
||||
predicateType,
|
||||
predicate,
|
||||
githubToken: 'gh-token',
|
||||
batchSize: 2,
|
||||
batchDelay: 500
|
||||
batchSize: 2
|
||||
}
|
||||
await main.run(inputs)
|
||||
|
||||
@@ -377,6 +375,54 @@ describe('action', () => {
|
||||
expect(scope.isDone()).toBe(true)
|
||||
})
|
||||
})
|
||||
|
||||
describe('when the subject count exceeds the max', () => {
|
||||
let dir = ''
|
||||
const filename = 'subject'
|
||||
|
||||
beforeEach(async () => {
|
||||
const subjectCount = 2501
|
||||
const content = 'file content'
|
||||
|
||||
// Set-up temp directory
|
||||
const tmpDir = await fs.realpath(os.tmpdir())
|
||||
dir = await fs.mkdtemp(tmpDir + path.sep)
|
||||
|
||||
// Add files for glob testing
|
||||
for (let i = 0; i < subjectCount; i++) {
|
||||
await fs.writeFile(path.join(dir, `${filename}-${i}`), content)
|
||||
}
|
||||
|
||||
// Set the GH context with private repository visibility and a repo owner.
|
||||
setGHContext({
|
||||
payload: { repository: { visibility: 'private' } },
|
||||
repo: { owner: 'foo', repo: 'bar' }
|
||||
})
|
||||
})
|
||||
|
||||
afterEach(async () => {
|
||||
// Clean-up temp directory
|
||||
await fs.rm(dir, { recursive: true })
|
||||
})
|
||||
|
||||
it('sets a failed status', async () => {
|
||||
const inputs: main.RunInputs = {
|
||||
...defaultInputs,
|
||||
subjectPath: path.join(dir, `${filename}-*`),
|
||||
predicateType,
|
||||
predicate,
|
||||
githubToken: 'gh-token'
|
||||
}
|
||||
await main.run(inputs)
|
||||
|
||||
expect(runMock).toHaveReturned()
|
||||
expect(setFailedMock).toHaveBeenCalledWith(
|
||||
new Error(
|
||||
'Too many subjects specified. The maximum number of subjects is 2500.'
|
||||
)
|
||||
)
|
||||
})
|
||||
})
|
||||
})
|
||||
|
||||
// Stubbing the GitHub context is a bit tricky. We need to use
|
||||
|
||||
@@ -151,7 +151,7 @@ describe('subjectFromInputs', () => {
|
||||
})
|
||||
})
|
||||
|
||||
describe('when the file eixts', () => {
|
||||
describe('when the file exists', () => {
|
||||
let dir = ''
|
||||
const filename = 'subject'
|
||||
const content = 'file content'
|
||||
|
||||
@@ -10,7 +10,7 @@ inputs:
|
||||
description: >
|
||||
Path to the artifact serving as the subject of the attestation. Must
|
||||
specify exactly one of "subject-path" or "subject-digest". May contain a
|
||||
glob pattern or list of paths (total subject count cannot exceed 64).
|
||||
glob pattern or list of paths (total subject count cannot exceed 2500).
|
||||
required: false
|
||||
subject-digest:
|
||||
description: >
|
||||
|
||||
58
dist/index.js
generated
vendored
58
dist/index.js
generated
vendored
@@ -80002,7 +80002,6 @@ Object.defineProperty(exports, "__esModule", ({ value: true }));
|
||||
const core = __importStar(__nccwpck_require__(42186));
|
||||
const main_1 = __nccwpck_require__(70399);
|
||||
const DEFAULT_BATCH_SIZE = 50;
|
||||
const DEFAULT_BATCH_DELAY = 5000;
|
||||
const inputs = {
|
||||
subjectPath: core.getInput('subject-path'),
|
||||
subjectName: core.getInput('subject-name'),
|
||||
@@ -80015,8 +80014,7 @@ const inputs = {
|
||||
// undocumented -- not part of public interface
|
||||
privateSigning: ['true', 'True', 'TRUE', '1'].includes(core.getInput('private-signing')),
|
||||
// internal only
|
||||
batchSize: DEFAULT_BATCH_SIZE,
|
||||
batchDelay: DEFAULT_BATCH_DELAY
|
||||
batchSize: DEFAULT_BATCH_SIZE
|
||||
};
|
||||
// eslint-disable-next-line @typescript-eslint/no-floating-promises
|
||||
(0, main_1.run)(inputs);
|
||||
@@ -80068,6 +80066,8 @@ const predicate_1 = __nccwpck_require__(72103);
|
||||
const style = __importStar(__nccwpck_require__(41583));
|
||||
const subject_1 = __nccwpck_require__(95206);
|
||||
const ATTESTATION_FILE_NAME = 'attestation.jsonl';
|
||||
const DELAY_INTERVAL_MS = 75;
|
||||
const DELAY_MAX_MS = 1200;
|
||||
/* istanbul ignore next */
|
||||
const logHandler = (level, ...args) => {
|
||||
// Send any HTTP-related log events to the GitHub Actions debug log
|
||||
@@ -80101,17 +80101,18 @@ async function run(inputs) {
|
||||
const outputPath = path_1.default.join(tempDir(), ATTESTATION_FILE_NAME);
|
||||
core.setOutput('bundle-path', outputPath);
|
||||
const subjectChunks = chunkArray(subjects, inputs.batchSize);
|
||||
let chunkCount = 0;
|
||||
// Generate attestations for each subject serially, working in batches
|
||||
for (const subjectChunk of subjectChunks) {
|
||||
// Delay between batches (only when chunkCount > 0)
|
||||
if (chunkCount++) {
|
||||
await new Promise(resolve => setTimeout(resolve, inputs.batchDelay));
|
||||
}
|
||||
for (let i = 0; i < subjectChunks.length; i++) {
|
||||
if (subjectChunks.length > 1) {
|
||||
core.info(`Processing subject batch ${chunkCount}/${subjectChunks.length}`);
|
||||
core.info(`Processing subject batch ${i + 1}/${subjectChunks.length}`);
|
||||
}
|
||||
for (const subject of subjectChunk) {
|
||||
// Calculate the delay time for this batch
|
||||
const delayTime = delay(i);
|
||||
for (const subject of subjectChunks[i]) {
|
||||
// Delay between attestations (only when chunk size > 1)
|
||||
if (i > 0) {
|
||||
await new Promise(resolve => setTimeout(resolve, delayTime));
|
||||
}
|
||||
const att = await (0, attest_1.createAttestation)(subject, predicate, {
|
||||
sigstoreInstance,
|
||||
pushToRegistry: inputs.pushToRegistry,
|
||||
@@ -80190,6 +80191,8 @@ const tempDir = () => {
|
||||
const chunkArray = (array, chunkSize) => {
|
||||
return Array.from({ length: Math.ceil(array.length / chunkSize) }, (_, index) => array.slice(index * chunkSize, (index + 1) * chunkSize));
|
||||
};
|
||||
// Calculate the delay time for a given iteration
|
||||
const delay = (iteration) => Math.min(DELAY_INTERVAL_MS * 2 ** iteration, DELAY_MAX_MS);
|
||||
const attestationURL = (id) => `${github.context.serverUrl}/${github.context.repo.owner}/${github.context.repo.repo}/attestations/${id}`;
|
||||
|
||||
|
||||
@@ -80287,6 +80290,7 @@ const crypto_1 = __importDefault(__nccwpck_require__(6113));
|
||||
const sync_1 = __nccwpck_require__(74393);
|
||||
const fs_1 = __importDefault(__nccwpck_require__(57147));
|
||||
const path_1 = __importDefault(__nccwpck_require__(71017));
|
||||
const MAX_SUBJECT_COUNT = 2500;
|
||||
const DIGEST_ALGORITHM = 'sha256';
|
||||
// Returns the subject specified by the action's inputs. The subject may be
|
||||
// specified as a path to a file or as a digest. If a path is provided, the
|
||||
@@ -80317,27 +80321,31 @@ exports.subjectFromInputs = subjectFromInputs;
|
||||
// Returns the subject specified by the path to a file. The file's digest is
|
||||
// calculated and returned along with the subject's name.
|
||||
const getSubjectFromPath = async (subjectPath, subjectName) => {
|
||||
const subjects = [];
|
||||
const digestedSubjects = [];
|
||||
const files = [];
|
||||
// Parse the list of subject paths
|
||||
const subjectPaths = parseList(subjectPath);
|
||||
// Expand the globbed paths to a list of files
|
||||
for (const subPath of subjectPaths) {
|
||||
// Expand the globbed path to a list of files
|
||||
/* eslint-disable-next-line github/no-then */
|
||||
const files = await glob.create(subPath).then(async (g) => g.glob());
|
||||
for (const file of files) {
|
||||
// Skip anything that is NOT a file
|
||||
if (!fs_1.default.statSync(file).isFile()) {
|
||||
continue;
|
||||
}
|
||||
const name = subjectName || path_1.default.parse(file).base;
|
||||
const digest = await digestFile(DIGEST_ALGORITHM, file);
|
||||
subjects.push({ name, digest: { [DIGEST_ALGORITHM]: digest } });
|
||||
}
|
||||
files.push(...(await glob.create(subPath).then(async (g) => g.glob())));
|
||||
}
|
||||
if (subjects.length === 0) {
|
||||
if (files.length > MAX_SUBJECT_COUNT) {
|
||||
throw new Error(`Too many subjects specified. The maximum number of subjects is ${MAX_SUBJECT_COUNT}.`);
|
||||
}
|
||||
for (const file of files) {
|
||||
// Skip anything that is NOT a file
|
||||
if (!fs_1.default.statSync(file).isFile()) {
|
||||
continue;
|
||||
}
|
||||
const name = subjectName || path_1.default.parse(file).base;
|
||||
const digest = await digestFile(DIGEST_ALGORITHM, file);
|
||||
digestedSubjects.push({ name, digest: { [DIGEST_ALGORITHM]: digest } });
|
||||
}
|
||||
if (digestedSubjects.length === 0) {
|
||||
throw new Error(`Could not find subject at path ${subjectPath}`);
|
||||
}
|
||||
return Promise.all(subjects);
|
||||
return digestedSubjects;
|
||||
};
|
||||
// Returns the subject specified by the digest of a file. The digest is returned
|
||||
// along with the subject's name.
|
||||
|
||||
@@ -5,7 +5,6 @@ import * as core from '@actions/core'
|
||||
import { run, RunInputs } from './main'
|
||||
|
||||
const DEFAULT_BATCH_SIZE = 50
|
||||
const DEFAULT_BATCH_DELAY = 5000
|
||||
|
||||
const inputs: RunInputs = {
|
||||
subjectPath: core.getInput('subject-path'),
|
||||
@@ -21,8 +20,7 @@ const inputs: RunInputs = {
|
||||
core.getInput('private-signing')
|
||||
),
|
||||
// internal only
|
||||
batchSize: DEFAULT_BATCH_SIZE,
|
||||
batchDelay: DEFAULT_BATCH_DELAY
|
||||
batchSize: DEFAULT_BATCH_SIZE
|
||||
}
|
||||
|
||||
// eslint-disable-next-line @typescript-eslint/no-floating-promises
|
||||
|
||||
29
src/main.ts
29
src/main.ts
@@ -10,6 +10,8 @@ import * as style from './style'
|
||||
import { SubjectInputs, subjectFromInputs } from './subject'
|
||||
|
||||
const ATTESTATION_FILE_NAME = 'attestation.jsonl'
|
||||
const DELAY_INTERVAL_MS = 75
|
||||
const DELAY_MAX_MS = 1200
|
||||
|
||||
export type RunInputs = SubjectInputs &
|
||||
PredicateInputs & {
|
||||
@@ -17,7 +19,6 @@ export type RunInputs = SubjectInputs &
|
||||
githubToken: string
|
||||
privateSigning: boolean
|
||||
batchSize: number
|
||||
batchDelay: number
|
||||
}
|
||||
|
||||
/* istanbul ignore next */
|
||||
@@ -62,22 +63,22 @@ export async function run(inputs: RunInputs): Promise<void> {
|
||||
core.setOutput('bundle-path', outputPath)
|
||||
|
||||
const subjectChunks = chunkArray(subjects, inputs.batchSize)
|
||||
let chunkCount = 0
|
||||
|
||||
// Generate attestations for each subject serially, working in batches
|
||||
for (const subjectChunk of subjectChunks) {
|
||||
// Delay between batches (only when chunkCount > 0)
|
||||
if (chunkCount++) {
|
||||
await new Promise(resolve => setTimeout(resolve, inputs.batchDelay))
|
||||
}
|
||||
|
||||
for (let i = 0; i < subjectChunks.length; i++) {
|
||||
if (subjectChunks.length > 1) {
|
||||
core.info(
|
||||
`Processing subject batch ${chunkCount}/${subjectChunks.length}`
|
||||
)
|
||||
core.info(`Processing subject batch ${i + 1}/${subjectChunks.length}`)
|
||||
}
|
||||
|
||||
for (const subject of subjectChunk) {
|
||||
// Calculate the delay time for this batch
|
||||
const delayTime = delay(i)
|
||||
|
||||
for (const subject of subjectChunks[i]) {
|
||||
// Delay between attestations (only when chunk size > 1)
|
||||
if (i > 0) {
|
||||
await new Promise(resolve => setTimeout(resolve, delayTime))
|
||||
}
|
||||
|
||||
const att = await createAttestation(subject, predicate, {
|
||||
sigstoreInstance,
|
||||
pushToRegistry: inputs.pushToRegistry,
|
||||
@@ -197,5 +198,9 @@ const chunkArray = <T>(array: T[], chunkSize: number): T[][] => {
|
||||
)
|
||||
}
|
||||
|
||||
// Calculate the delay time for a given iteration
|
||||
const delay = (iteration: number): number =>
|
||||
Math.min(DELAY_INTERVAL_MS * 2 ** iteration, DELAY_MAX_MS)
|
||||
|
||||
const attestationURL = (id: string): string =>
|
||||
`${github.context.serverUrl}/${github.context.repo.owner}/${github.context.repo.repo}/attestations/${id}`
|
||||
|
||||
@@ -6,6 +6,7 @@ import path from 'path'
|
||||
|
||||
import type { Subject } from '@actions/attest'
|
||||
|
||||
const MAX_SUBJECT_COUNT = 2500
|
||||
const DIGEST_ALGORITHM = 'sha256'
|
||||
|
||||
export type SubjectInputs = {
|
||||
@@ -54,34 +55,41 @@ const getSubjectFromPath = async (
|
||||
subjectPath: string,
|
||||
subjectName?: string
|
||||
): Promise<Subject[]> => {
|
||||
const subjects: Subject[] = []
|
||||
const digestedSubjects: Subject[] = []
|
||||
const files: string[] = []
|
||||
|
||||
// Parse the list of subject paths
|
||||
const subjectPaths = parseList(subjectPath)
|
||||
|
||||
// Expand the globbed paths to a list of files
|
||||
for (const subPath of subjectPaths) {
|
||||
// Expand the globbed path to a list of files
|
||||
/* eslint-disable-next-line github/no-then */
|
||||
const files = await glob.create(subPath).then(async g => g.glob())
|
||||
|
||||
for (const file of files) {
|
||||
// Skip anything that is NOT a file
|
||||
if (!fs.statSync(file).isFile()) {
|
||||
continue
|
||||
}
|
||||
|
||||
const name = subjectName || path.parse(file).base
|
||||
const digest = await digestFile(DIGEST_ALGORITHM, file)
|
||||
|
||||
subjects.push({ name, digest: { [DIGEST_ALGORITHM]: digest } })
|
||||
}
|
||||
files.push(...(await glob.create(subPath).then(async g => g.glob())))
|
||||
}
|
||||
|
||||
if (subjects.length === 0) {
|
||||
if (files.length > MAX_SUBJECT_COUNT) {
|
||||
throw new Error(
|
||||
`Too many subjects specified. The maximum number of subjects is ${MAX_SUBJECT_COUNT}.`
|
||||
)
|
||||
}
|
||||
|
||||
for (const file of files) {
|
||||
// Skip anything that is NOT a file
|
||||
if (!fs.statSync(file).isFile()) {
|
||||
continue
|
||||
}
|
||||
|
||||
const name = subjectName || path.parse(file).base
|
||||
const digest = await digestFile(DIGEST_ALGORITHM, file)
|
||||
|
||||
digestedSubjects.push({ name, digest: { [DIGEST_ALGORITHM]: digest } })
|
||||
}
|
||||
|
||||
if (digestedSubjects.length === 0) {
|
||||
throw new Error(`Could not find subject at path ${subjectPath}`)
|
||||
}
|
||||
|
||||
return Promise.all(subjects)
|
||||
return digestedSubjects
|
||||
}
|
||||
|
||||
// Returns the subject specified by the digest of a file. The digest is returned
|
||||
|
||||
Reference in New Issue
Block a user