You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This is a collection of issues related to Module execution. In particular these have been raised when running orchestrator:keynesian_beauty_contest, other modules, possibly orchestrators, might throw the errors errors as well.
Investigation
On a local (darwin/arm64) set up:
Node logs tend to be not informative at all. They show the same error line displayed by the SDK.
I usedhtop to keep an eye on resource usage, it doesn't look like it goes OOM
Note that another non-trivial orch. (sometimes) errors out - retry --until=fail naptha run orchestrator:multiagent_chat -p "prompt='i would like to count up to ten, one number at a time. ill start. one.'" --agent_nodes "localhost,localhost" --kb_nodes "localhost".
Is the error limited to (keynesian) orchestrators?
Simple agent doesn't error out - retry --until=fail naptha run agent:hello_world_agent -p "firstname=sam surname=altman"
Solution
Increasing consumer_timeout in main.py seems to have a positive impact (with 10 agents). Still testing it.
The text was updated successfully, but these errors were encountered:
Description
This is a collection of issues related to Module execution. In particular these have been raised when running
orchestrator:keynesian_beauty_contest
, other modules, possibly orchestrators, might throw the errors errors as well.Investigation
On a local (darwin/arm64) set up:
htop
to keep an eye on resource usage, it doesn't look like it goes OOMNote that another non-trivial orch. (sometimes) errors out -
retry --until=fail naptha run orchestrator:multiagent_chat -p "prompt='i would like to count up to ten, one number at a time. ill start. one.'" --agent_nodes "localhost,localhost" --kb_nodes "localhost"
.Is the error limited to (keynesian) orchestrators?
retry --until=fail naptha run agent:hello_world_agent -p "firstname=sam surname=altman"
Solution
Increasing
consumer_timeout
inmain.py
seems to have a positive impact (with 10 agents). Still testing it.The text was updated successfully, but these errors were encountered: