Page MenuHomePhabricator

[services] Tunnelbroker - Skipping session checks in a stream when `skip_authentication` flag is set
ClosedPublic

Authored by max on Jan 12 2023, 8:05 AM.
Tags
None
Referenced Files
F3485306: D6253.id21752.diff
Tue, Dec 17, 7:55 PM
F3485305: D6253.id21750.diff
Tue, Dec 17, 7:55 PM
F3485304: D6253.id20874.diff
Tue, Dec 17, 7:55 PM
F3485291: D6253.id.diff
Tue, Dec 17, 7:54 PM
F3485283: D6253.diff
Tue, Dec 17, 7:52 PM
Unknown Object (File)
Mon, Dec 9, 9:46 AM
Unknown Object (File)
Nov 15 2024, 7:59 AM
Unknown Object (File)
Nov 15 2024, 6:25 AM
Subscribers

Details

Summary

This diff introduces changes to the bidirectional stream gRPC handler to skip session authentication checks by adding the sessions.skip_authentication config flag.
In case of the flag was provided we are skipping getting the session parameters from the database and expecting the device will be provided as metadata instead. Also, we are creating an empty SessionItem instance to provide compatibility with the following code in the handler. We are adding skipping of the changing of "Online" status for the session in case the flag is provided.

Linear task: ENG-2642

Test Plan
  1. Add the skip_authentication flag to the [sessions] section of the config file.
  2. Connect to the bidirectional messages stream by providing the deviceID parameter instead of sessionID.

The expected result is messages are delivered to the certain deviceID without prior session creation and authentication.

Diff Detail

Repository
rCOMM Comm
Lint
Lint Not Applicable
Unit
Tests Not Applicable

Event Timeline

max held this revision as a draft.
max published this revision for review.Jan 12 2023, 8:21 AM
max edited the summary of this revision. (Show Details)
max edited the test plan for this revision. (Show Details)
max added a reviewer: bartek. max added 1 blocking reviewer(s): jon.

otherwise looks good to me

services/tunnelbroker/src/server/mod.rs
138 ↗(On Diff #20874)

Outside the context of this diff, but SessionItem sounds like a "one of many" object to me. SessionContext would be more accurate in describing its purpose, which is to hold some metadata about the current session.

This revision is now accepted and ready to land.Jan 12 2023, 11:24 AM
This revision now requires review to proceed.Jan 12 2023, 11:24 AM
bartek added inline comments.
services/tunnelbroker/src/server/mod.rs
138–146 ↗(On Diff #20874)

Nit: I'd extract this to an initializer function, sth like SessionItem::new(device_id: String)

This revision is now accepted and ready to land.Jan 12 2023, 11:42 AM

Rebasing on the parent changes.

max added inline comments.
services/tunnelbroker/src/server/mod.rs
138 ↗(On Diff #20874)

Outside the context of this diff, but SessionItem sounds like a "one of many" object to me. SessionContext would be more accurate in describing its purpose, which is to hold some metadata about the current session.

That name came from the legacy C++ codebase and refactoring it seems doesn't makes sense now as we possibly switch to Rust totally.

138–146 ↗(On Diff #20874)

Nit: I'd extract this to an initializer function, sth like SessionItem::new(device_id: String)

Thanks, @bartek! But this part is a single usage (for our testing Tunnelbroker usage as Redis replacement only) and adding the initializer for that single usage seems maybe overengineering to me.