Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

fix: increase max packet size for built-in bridge #3059

Merged
merged 4 commits into from
Aug 13, 2024
Merged

fix: increase max packet size for built-in bridge #3059

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
359384e
Increase max packet size for MQTT bridge (#3056)
jarhodes314 Aug 8, 2024
183aba8
Simplify logic for handling QoS 0 messages
jarhodes314 Aug 8, 2024
10b4d65
Document built-in bridge configuration
jarhodes314 Aug 8, 2024
5cfb663
Increase max message size to 256MiB
jarhodes314 Aug 13, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions crates/common/tedge_config/src/tedge_config_cli/tedge_config.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -769,6 +769,9 @@ define_tedge_config! {

bridge: {
#[tedge_config(default(value = false))]
#[tedge_config(example = "false")]
#[tedge_config(note = "After changing this value, run `tedge reconnect <cloud>` to apply the changes")]
/// Enables the built-in bridge when running tedge-mapper
built_in: bool,

reconnect_policy: {
Expand Down
28 changes: 10 additions & 18 deletions crates/extensions/tedge_mqtt_bridge/src/lib.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,6 +57,8 @@ use crate::topics::matches_ignore_dollar_prefix;
use crate::topics::TopicConverter;
pub use config::*;

const MAX_PACKET_SIZE: usize = 10 * 1024 * 1024;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks like we have some inconsistency with packet sizes used across the project.

  1. In mqtt_channel::config::Config: 1024 * 1024
  2. In bridge::common_mosquitto_config::CommonMosquittoConfig: 256 * 1024 * 1024 - 1
  3. In tedge::cli::mqtt::MAX_PACKET_SIZE: 10 * 1024 * 1024

Any reason why you chose the 3rd option and not the second one, keeping it consistent with the existing bridge config?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I didn't realise that mosquitto used a completely different name for the configuration, so didn't notice it. I'll change this to be consistent with the existing bridge.


pub struct MqttBridgeActorBuilder {}

impl MqttBridgeActorBuilder {
Expand Down Expand Up @@ -90,6 +92,7 @@ impl MqttBridgeActorBuilder {
if let Some(tls_config) = local_tls_config {
local_config.set_transport(Transport::tls_with_config(tls_config.into()));
}
local_config.set_max_packet_size(MAX_PACKET_SIZE, MAX_PACKET_SIZE);
local_config.set_manual_acks(true);
local_config.set_last_will(LastWill::new(
&health_topic.name,
Expand All @@ -102,6 +105,7 @@ impl MqttBridgeActorBuilder {
let reconnect_policy = tedge_config.mqtt.bridge.reconnect_policy.clone();

cloud_config.set_manual_acks(true);
cloud_config.set_max_packet_size(MAX_PACKET_SIZE, MAX_PACKET_SIZE);

let (local_client, local_event_loop) = AsyncClient::new(local_config, 10);
let (cloud_client, cloud_event_loop) = AsyncClient::new(cloud_config, 10);
Expand Down Expand Up @@ -377,28 +381,16 @@ async fn half_bridge(

// Keep track of packet IDs so we can acknowledge messages
Event::Outgoing(Outgoing::Publish(pkid)) => {
if pkid == 0 {
// Messages with pkid 0 (meaning QoS=0) should not be added to the hashmap
// as multiple messages with the pkid=0 can be received
match companion_bridge_half.recv().await {
// A message was forwarded by the other bridge half, note the packet id
Some(Some((topic, msg))) => {
loop_breaker.forward_on_topic(topic, &msg);
}

// A healthcheck message was published, ignore this packet id
Some(None) => {}

// The other bridge half has disconnected, break the loop and shut down the bridge
None => break,
}
} else if let hash_map::Entry::Vacant(e) = forward_pkid_to_received_msg.entry(pkid)
{
if let hash_map::Entry::Vacant(e) = forward_pkid_to_received_msg.entry(pkid) {
match companion_bridge_half.recv().await {
// A message was forwarded by the other bridge half, note the packet id
Some(Some((topic, msg))) => {
loop_breaker.forward_on_topic(topic, &msg);
e.insert(msg);
if pkid != 0 {
// Messages with pkid 0 (meaning QoS=0) should not be added to the hashmap
// as multiple messages with the pkid=0 can be received
e.insert(msg);
}
Comment on lines +389 to +393
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can this pkid = 0 case be covered with a unit test?

Copy link
Contributor Author

@jarhodes314 jarhodes314 Aug 9, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Not easily. In an ideal world, we would have some deterministic tests where we control the broker we connect to directly, so we can send QoS 0 messages. This would also shore up the existing integration tests, which have to flood the bridge with thousands of messages before disconnecting to reproduce the desired behaviour.

I think this might be worth investigating in a time-boxed manner, but I'll create a separate ticket for it.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can this pkid = 0 case be covered with a unit test?

This is tested by the system test "Support publishing QoS 0 messages to c8y topic #2960".

Sure, this test is not deterministic (as messages with QoS 0 might not be delivered), but I think that is enough.

}

// A healthcheck message was published, ignore this packet id
Expand Down