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
Currently the memory setup means on the "large" node we end up too memory constrained which can cause contention issues and panics.
We need to benchmark the services and ensure we allocate sufficient memory to the different systems.
Crucially we also need to set a "mem_limit" on each container to ensure that we can avoid noisy neighbor scenarios -> CPU constraints remain but at least memory won't be in competition anymore which should improve stability.
The text was updated successfully, but these errors were encountered:
Currently the memory setup means on the "large" node we end up too memory constrained which can cause contention issues and panics.
We need to benchmark the services and ensure we allocate sufficient memory to the different systems.
Crucially we also need to set a "mem_limit" on each container to ensure that we can avoid noisy neighbor scenarios -> CPU constraints remain but at least memory won't be in competition anymore which should improve stability.
The text was updated successfully, but these errors were encountered: