I am developing a real-time conversational agent application using the Google ADK agent.
Because the default configuration is subject to a 10-minute connection time limit, I enabled session resumption according to the official documentation:
[google github io/adk-docs streaming/dev-guide/part4/ live-api-session-resumption] (google github io/adk-docs streaming/dev-guide/part4/ live-api-session-resumption)
However, during testing, when the session reaches the 10-minute mark, I consistently receive the following error:
received 1011 (internal error) Deadline expired before operation could complete.; then sent 1011 (internal error) Deadline expired before operation could complete.
As a result, the connection is terminated.
According to the official documentation:
“ADK automates this entirely: When you enable session resumption in RunConfig, ADK automatically handles all reconnection logic—detecting connection closures, caching resumption handles, and reconnecting seamlessly in the background. You don’t need to write any reconnection code. Sessions continue seamlessly beyond the 10-minute connection limit, handling connection timeouts, network disruptions, and planned reconnections automatically.”
Based on this description, I believe no additional handling should be required on my side. What I need to do is to configure the following in run_config:
run_config = RunConfig(
session_resumption=types.SessionResumptionConfig()
)
Unfortunately, this does not work as expected.
The model I am using is gemini-2.5-flash-native-audio-preview-12-2025 via the Gemini API.
ADK Version: 1.22.0
I also tested Vertex AI–based models, including gemini-live-2.5-flash-native-audio and gemini-live-2.5-flash-preview-native-audio-09-2025, and they work correctly with session resumption enabled. However, based on my evaluation, gemini-2.5-flash-native-audio-preview-12-2025 provides better overall performance, but it is not available on Vertex AI.