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
Is there any change to get option to limit pruntime utilization in benchmark stage?
Cuz it sometimes makes unable to access runtime server or even entire machine.
I understand idea at current stage, but let say reducing utilization to 90% by adding some sleep period in benchmark loop would left some resources for server - so PRB could freely access it.
I would prefer reducing benchmark process to 90-95% rather than leaving 1 core for that purpose.
EDIT:
Interesting thing I discovered which seems to confirm need to limit utilization - I connected into worker machine and tried simplest endpoint curl localhost:8000/get_info
Sometimes it takes really long to get response (I have noticed even 160s).
The text was updated successfully, but these errors were encountered:
Is there any change to get option to limit pruntime utilization in benchmark stage?
Cuz it sometimes makes unable to access runtime server or even entire machine.
I understand idea at current stage, but let say reducing utilization to 90% by adding some sleep period in benchmark loop would left some resources for server - so PRB could freely access it.
I would prefer reducing benchmark process to 90-95% rather than leaving 1 core for that purpose.
EDIT:
Interesting thing I discovered which seems to confirm need to limit utilization - I connected into worker machine and tried simplest endpoint
curl localhost:8000/get_info
Sometimes it takes really long to get response (I have noticed even 160s).
The text was updated successfully, but these errors were encountered: