"Rate limit exceeded" when sending a single message

[lushing]

When using the Manager's send_message method, I will always run into the following error:

[INFO  libsignal_service::sender] establishing new session with ProtocolAddress { name: "XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX", device_id: 1 }
[WARN  presage::config::sled] trusting new identity ProtocolAddress { name: "XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX", device_id: 1 }
[INFO  libsignal_protocol::session] set_unacknowledged_pre_key_message for: XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX.1 with preKeyId: XXXXXXXX
[INFO  libsignal_protocol::state::session] Skipping archive, current session state is fresh
[INFO  libsignal_protocol::session] set_unacknowledged_pre_key_message for: XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX.2 with preKeyId: XXX
[INFO  libsignal_protocol::session_cipher] Building PreKeyWhisperMessage for: XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX.1 with preKeyId: XXXXXXXX
[INFO  libsignal_protocol::session] set_unacknowledged_pre_key_message for: XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX.2 with preKeyId: XXX
[INFO  libsignal_protocol::session_cipher] Building PreKeyWhisperMessage for: XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX.1 with preKeyId: XXXXXXXX
[INFO  libsignal_protocol::session] set_unacknowledged_pre_key_message for: XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX.2 with preKeyId: XXX
[INFO  libsignal_protocol::session_cipher] Building PreKeyWhisperMessage for: XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX.1 with preKeyId: XXXXXXXX
[INFO  libsignal_protocol::session] set_unacknowledged_pre_key_message for: XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX.2 with preKeyId: XXX
[INFO  libsignal_protocol::session_cipher] Building PreKeyWhisperMessage for: XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX.1 with preKeyId: XXXXXXXX
Error: libsignal-service sending error: Rate limit exceeded

Caused by:
    0: Rate limit exceeded
    1: Rate limit exceeded

This occurs whenever I try to send only one message to another Signal account. I've tried many variations of the code, but always with the same result.

Here is the relevant code:

let number = phonenumber::parse(None, "+17607067425")?;

let timestamp = std::time::SystemTime::now()
    .duration_since(UNIX_EPOCH)
    .expect("negative time")
    .as_millis() as u64;

let message = ContentBody::DataMessage(DataMessage {
    body: Some("test"),
    timestamp: Some(timestamp),
    ..Default::default()
});

manager.send_message(number, message, timestamp).await?;

[boxdot]

I am using this method in Gurk without any problems. It might be that you linked/unlinked a device too often in a short period of time, which leads to rate limiting. There is also a simple whoami method that gives you your ID back. Maybe you can try this one to see if it is also rate-limited.

[gferon]

I discovered a bug but I committed a fix sneakily as part of #20, sorry for that. I'll try to be more transparent in the future, if that can help!

The problem is something we've already had to deal with in the past, and is related to the way we store sessions.

The switch to the Rust version of libsignal-protocol re-introduced the bug because the std::fmt::Display implementation of ProtocolAddress writes {phone_number}.{device_id} and we were expecting {phone_number}-{device_id} in the implementation of StoreSessionExt::get_sub_devices_sessions.

[gferon]

Forgot to ask: @lushing could you try again with the latest commit from main and report back here?

[lushing]

@gferon This has indeed fixed the bug, and sending messages now works as expected. Thanks!