feat(windows): move connlib into a subprocess #3349

ReactorScram · 2024-01-23T00:29:55Z

Todo

Give feedback

Refactor

Refactor
Options
Successfully updated the issue's project

There was an error updating the issue's project
Research tokio_codec, tokio_util, StreamExt, and splitting the named pipe into read and write halves

Research tokio_codec, tokio_util, StreamExt, and splitting the named pipe into read and write halves
Options
Successfully updated the issue's project

There was an error updating the issue's project
Document security of named pipes (e.g. can non-root processes read them?)

Document security of named pipes (e.g. can non-root processes read them?)
Options
Successfully updated the issue's project

There was an error updating the issue's project
Document no-harness tests for cargo test

Document no-harness tests for cargo test
Options
Successfully updated the issue's project

There was an error updating the issue's project
Fix log files

Fix log files
Options
Successfully updated the issue's project

There was an error updating the issue's project
Make connlib subprocess handle token instead of GUI process

Make connlib subprocess handle token instead of GUI process
Options
Successfully updated the issue's project

There was an error updating the issue's project
Negative test for security features

Negative test for security features
Options
Successfully updated the issue's project

There was an error updating the issue's project
Options

Tests

Give feedback

Make sure the log files are separated (I assume two processes won't automatically cooperate to write to the same file correctly)

Make sure the log files are separated (I assume two processes won't automatically cooperate to write to the same file correctly)
Options
Successfully updated the issue's project

There was an error updating the issue's project
Test export log zip

Test export log zip
Options
Successfully updated the issue's project

There was an error updating the issue's project
Make sure the crash handler does something reasonable for either process crashing

Make sure the crash handler does something reasonable for either process crashing
Options
Successfully updated the issue's project

There was an error updating the issue's project
Make sure signing out closes the tunnel process gracefully

Make sure signing out closes the tunnel process gracefully
Options
Successfully updated the issue's project

There was an error updating the issue's project
Make sure quitting the GUI closes the tunnel process gracefully

Make sure quitting the GUI closes the tunnel process gracefully
Options
Successfully updated the issue's project

There was an error updating the issue's project
Make sure killing the GUI process kills the tunnel process

Make sure killing the GUI process kills the tunnel process
Options
Successfully updated the issue's project

There was an error updating the issue's project
Make sure killing the tunnel process causes the GUI to do something reasonable

Make sure killing the tunnel process causes the GUI to do something reasonable
Options
Successfully updated the issue's project

There was an error updating the issue's project
Options

…est'

… and make sure normal execution doesn't depend on debug subcommands

…rther from production subcommands

…ram/ipc-test

…-subcommands

…factor-debug-subcommands

…ram/ipc-test

…leak

…-debug-subcommands

…ram/ipc-test

Signed-off-by: Reactor Scram <ReactorScram@users.noreply.github.com>

…leak

…neater

…eactorscram/ipc-connlib

…cess can work

I heard of some OS where a child process could inherit its parent's ID. Maybe I dreamt it.

… by half

…factor Controller

ReactorScram · 2024-01-26T15:45:57Z

I tried to refactor with this too but it was taking too long and I hit a snag where I wanted to read the cookie before the codec was created https://docs.rs/tokio-util/latest/tokio_util/codec/length_delimited/struct.LengthDelimitedCodec.html#

…for testing

ReactorScram · 2024-01-26T18:26:52Z

Per standup, will abandon this and just split off a subprocess crate, since it took a long time to figure out and some other Rust + Windows project might find it convenient.

ReactorScram · 2024-01-26T21:53:15Z

the crate was split out in #3414, if that merges and firezone-connection works well, we can close this

ReactorScram · 2024-01-29T17:30:30Z

Closing, if we need this again subzone has moved over here https://github.com/firezone/subzone

ReactorScram added 30 commits January 17, 2024 19:31

wip(windows): IPC mock

d659577

wip(windows): IPC mock

8ff6a68

Merge remote-tracking branch 'origin/main' into reactorscram/ipc-test

390c38c

wip(windows): try out ipc-channel crate

30950fc

wip(windows): add timeouts and stuff

fce1c57

wip(windows): replace ipc-channel crate with Tokio's named_pipe module

88ed540

wip(windows): upgrade from u8 to full enum messages

a30f11f

wip(windows): add a subcommand for tests that don't fit into 'cargo t…

5c08b7a

…est'

fix(windows): typo

27c05bc

Merge remote-tracking branch 'origin/main' into reactorscram/ipc-test

ee5e424

Merge remote-tracking branch 'origin/main' into reactorscram/ipc-test

91e5a0e

refactor(windows): move CLI things inside client.rs

7eac132

refactor(windows): remove unused subcommands and redundant arguments,…

1fc8656

… and make sure normal execution doesn't depend on debug subcommands

refactor(windows): move debug subcommands closer to their code and fu…

ba432bf

…rther from production subcommands

Merge branch 'reactorscram/refactor-debug-subcommands' into reactorsc…

fe8d358

…ram/ipc-test

wip(windows): multi-process test for IPC

5f21c2c

wip(windows): make Windows kill child processes when the parent exits

3f6a295

Merge remote-tracking branch 'origin/main' into reactorscram/refactor…

831ec3f

…-subcommands

Merge branch 'reactorscram/refactor-subcommands' into reactorscram/re…

37d888c

…factor-debug-subcommands

Merge branch 'reactorscram/refactor-debug-subcommands' into reactorsc…

06dff39

…ram/ipc-test

Merge branch 'reactorscram/ipc-test' into reactorscram/ipc-crash-test

1145279

Merge branch 'reactorscram/ipc-crash-test' into reactorscram/process-…

47288d3

…leak

Merge remote-tracking branch 'origin/main' into reactorscram/refactor…

1fae3f6

…-debug-subcommands

Merge branch 'main' into reactorscram/refactor-debug-subcommands

5de44d1

Merge branch 'reactorscram/refactor-debug-subcommands' into reactorsc…

e4692ac

…ram/ipc-test

Merge branch 'reactorscram/refactor-debug-subcommands' into reactorsc…

4afd0e2

…ram/ipc-test

Merge branch 'reactorscram/ipc-test' into reactorscram/ipc-crash-test

dd8c1c9

Signed-off-by: Reactor Scram <ReactorScram@users.noreply.github.com>

Merge branch 'reactorscram/ipc-crash-test' into reactorscram/process-…

fe64c47

…leak

refactor(windows): move the IPC test code into the IPC module, seems …

a1a31a4

…neater

ci(windows): trigger the multi-process IPC test during CI runs

b260a40

ReactorScram added 7 commits January 23, 2024 14:58

wip(windows): named pipes bug

cb371a8

wip(windows): it's getting closer

15dbd27

Merge remote-tracking branch 'origin/main' into reactorscram/ipc-api

7bfddb2

Merge branch 'main' into reactorscram/ipc-api

1aeea0b

Merge branch 'reactorscram/ipc-api' into reactorscram/ipc-connlib

a9e97a3

Merge branch 'main' into reactorscram/ipc-api

199273c

Merge branch 'reactorscram/ipc-api' into reactorscram/ipc-connlib

898920e

ReactorScram added the area/windows_client Issues related to the Windows client label Jan 24, 2024

ReactorScram added 3 commits January 24, 2024 12:55

Merge remote-tracking branch 'origin/reactorscram/ipc-connlib' into r…

1feca4e

…eactorscram/ipc-connlib

wip(windows IPC): fix numerous bugs in IPC so that the connlib subpro…

b4e6fba

…cess can work

test(windows): make sure processes really have unique IDs

262d654

I heard of some OS where a child process could inherit its parent's ID. Maybe I dreamt it.

ReactorScram mentioned this pull request Jan 25, 2024

refactor(windows): simpler API for spawning subprocesses #3347

Closed

ReactorScram added 5 commits January 25, 2024 09:23

wip(windows): this one works

7531077

Merge remote-tracking branch 'origin/main' into reactorscram/ipc-connlib

3c6a77b

fix(windows): fix graceful shutdown and other IPC bugs

aacc07a

fix(windows): use tokio's process module to remove some sleeps

cec0d5c

Merge remote-tracking branch 'origin/main' into reactorscram/ipc-connlib

a0c990f

ReactorScram changed the base branch from reactorscram/ipc-api to main January 25, 2024 18:18

ReactorScram added 5 commits January 25, 2024 13:44

refactor(windows): apply feedback from Thomas

a3a8a32

docs(windows): comments

1bdd5a9

refactor(windows): split up IPC into sub-modules

c83865d

refactor(windows): use 'tokio::io::split' to reduce 'client.rs' lines…

4ce083b

… by half

refactor(windows): use 'tokio::io::split' for the server too, also re…

de87c11

…factor Controller

ReactorScram added 2 commits January 26, 2024 11:11

fix(windows): separate log files between the two processes

f49af01

fix(windows): fix log export and count and add flag to crash connlib …

8f87069

…for testing

ReactorScram closed this Jan 29, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat(windows): move connlib into a subprocess #3349

feat(windows): move connlib into a subprocess #3349

ReactorScram commented Jan 23, 2024 •

edited

Loading

Todo

Tests

ReactorScram commented Jan 26, 2024

ReactorScram commented Jan 26, 2024

ReactorScram commented Jan 26, 2024

ReactorScram commented Jan 29, 2024

feat(windows): move connlib into a subprocess #3349

feat(windows): move connlib into a subprocess #3349

Conversation

ReactorScram commented Jan 23, 2024 • edited Loading

Todo

Tests

ReactorScram commented Jan 26, 2024

ReactorScram commented Jan 26, 2024

ReactorScram commented Jan 26, 2024

ReactorScram commented Jan 29, 2024

ReactorScram commented Jan 23, 2024 •

edited

Loading