The motivation for supporting an alternative to WebSockets is that while all browsers supported by Grist offer native WebSocket support, some networking environments do not allow WebSocket traffic.
Engine.IO is used as the underlying implementation of HTTP long polling. The Grist client will first attempt a regular WebSocket connection, using the same protocol and endpoints as before, but fall back to long polling using Engine.IO if the WebSocket connection fails.
Include these changes:
- CORS websocket requests are now rejected as a stronger security measure. This shouldn’t affect anything in practice; but previously it could be possible to make unauthenticated websocket requests from another origin.
- GRIST_HOST variable no longer affects CORS responses (also should not affect anything in practice, as it wasn't serving a useful purpose)
* Fix server crash when client passes malformed JSON
* Take remarks into account
---------
Co-authored-by: Florent FAYOLLE <florent.fayolle@beta.gouv.fr>
Summary:
- Node has a strong recommendation to assume bad state and exit promptly on
unhandled exceptions and rejections. We follow it, and only make an effort to
clean up before exiting, and to log the error in a more standard way.
- The only case seen in recent month of an unhandled rejection was for
attempting to write overly large JSON to a Client websocket. Ensure that's
handled, and add a test case that artificially reproduces this scenario.
Test Plan:
Added a test case for failing write to Client, and a test case that unhandled
errors indeed kill the server but with an attempt at cleanup.
Reviewers: georgegevoian
Reviewed By: georgegevoian
Differential Revision: https://phab.getgrist.com/D4124
Summary:
Changes the minimum version of Node to 18, and updates the Docker images and GitHub workflows to build Grist with Node 18.
Also updates various dependencies and scripts to support building running tests with arm64 builds of Node.
Test Plan: Existing tests.
Reviewers: paulfitz
Reviewed By: paulfitz
Differential Revision: https://phab.getgrist.com/D3968
Summary:
- /status accepts new optional query parameters: db=1, redis=1, and timeout=<ms> (defaults to 10_000).
- These verify that the server can make trivial calls to DB/Redis, and that they return within the timeout.
- New HealthCheck tests simulates DB and Redis problems.
- Added resilience to Redis reconnects (helped by a test case that simulates disconnects)
- When closing Redis-based session store, disconnect from Redis (to avoid hanging tests)
Some associated test reorg:
- Move stripeTools out of test/nbrowser, and remove an unnecessary dependency,
to avoid starting up browser for gen-server tests.
- Move TcpForwarder to its own file, to use in the new test.
Test Plan: Added a new HealthCheck test that simulates DB and Redis problems.
Reviewers: georgegevoian
Reviewed By: georgegevoian
Differential Revision: https://phab.getgrist.com/D4054
Summary:
- Implements MemoryPool for waiting on memory reservations.
- Uses MemoryPool to control memory used for stringifying JSON responses in Client.ts
- Limits total size of _missedMessages that may be queued for a particular client.
- Upgrades ws library, which may reduce memory usage, and allows pausing the websocket for testing.
- The upgrade changed subtle behavior corners, requiring various fixes to code and tests.
- dos.ts:
- Includes Paul's fixes and updates to the dos.ts script for manual stress-testing.
- Logging tweaks, to avoid excessive dumps on uncaughtError, and include timestamps.
Test Plan:
- Includes a test that measures heap size, and fails without memory management.
- Includes a unittest for MemoryPool
- Some cleanup and additions to TestServer helper; in particular adds makeUserApi() helper used in multiple tests.
- Some fixes related to ws upgrade.
Reviewers: paulfitz
Reviewed By: paulfitz
Differential Revision: https://phab.getgrist.com/D3974
Summary:
Building:
- Builds no longer wait for tsc for either client, server, or test targets. All use esbuild which is very fast.
- Build still runs tsc, but only to report errors. This may be turned off with `SKIP_TSC=1` env var.
- Grist-core continues to build using tsc.
- Esbuild requires ES6 module semantics. Typescript's esModuleInterop is turned
on, so that tsc accepts and enforces correct usage.
- Client-side code is watched and bundled by webpack as before (using esbuild-loader)
Code changes:
- Imports must now follow ES6 semantics: `import * as X from ...` produces a
module object; to import functions or class instances, use `import X from ...`.
- Everything is now built with isolatedModules flag. Some exports were updated for it.
Packages:
- Upgraded browserify dependency, and related packages (used for the distribution-building step).
- Building the distribution now uses esbuild's minification. babel-minify is no longer used.
Test Plan: Should have no behavior changes, existing tests should pass, and docker image should build too.
Reviewers: georgegevoian
Reviewed By: georgegevoian
Subscribers: alexmojaki
Differential Revision: https://phab.getgrist.com/D3506
Summary:
- Upgrades to build-related packages:
- Upgrade typescript, related libraries and typings.
- Upgrade webpack, eslint; add tsc-watch, node-dev, eslint_d.
- Build organization changes:
- Build webpack from original typescript, transpiling only; with errors still
reported by a background tsc watching process.
- Typescript-related changes:
- Reduce imports of AWS dependencies (very noticeable speedup)
- Avoid auto-loading global @types
- Client code is now built with isolatedModules flag (for safe transpilation)
- Use allowJs to avoid copying JS files manually.
- Linting changes
- Enhance Arcanist ESLintLinter to run before/after commands, and set up to use eslint_d
- Update eslint config, and include .eslintignore to avoid linting generated files.
- Include a bunch of eslint-prompted and eslint-generated fixes
- Add no-unused-expression rule to eslint, and fix a few warnings about it
- Other items:
- Refactor cssInput to avoid circular dependency
- Remove a bit of unused code, libraries, dependencies
Test Plan: No behavior changes, all existing tests pass. There are 30 tests fewer reported because `test_gpath.py` was removed (it's been unused for years)
Reviewers: paulfitz
Reviewed By: paulfitz
Subscribers: paulfitz
Differential Revision: https://phab.getgrist.com/D3498
Summary:
- Substantial refactoring of the logic when the server fails to send some
messages to a client.
- Add seqId numbers to server messages to ensure reliable order.
- Add a needReload flag in clientConnect for a clear indication whent the
browser client needs to reload the app.
- Reproduce some potential failure scenarios in a test case (some of which
previously could have led to incorrectly ordered messages).
- Convert other Comm tests to typescript.
- Tweak logging of Comm and Client to be slightly more concise (in particular,
avoid logging sessionId)
Note that despite the big refactoring, this only addresses a fairly rare
situation, with websocket failures while server is trying to send to the
client. It includes no improvements for failures while the client is sending to
the server.
(I looked for an existing library that would take care of these issues. A relevant article I found is https://docs.microsoft.com/en-us/azure/azure-web-pubsub/howto-develop-reliable-clients, but it doesn't include a library for both ends, and is still in review. Other libraries with similar purposes did not inspire enough confidence.)
Test Plan: New test cases, which reproduce some previously problematic scenarios.
Reviewers: paulfitz
Reviewed By: paulfitz
Differential Revision: https://phab.getgrist.com/D3470
