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
We're running taskq workers on kubernetes pods, but have encountered a less than ideal situation with the hasFreeSystemResources check. It appears that the Sysinfo checks are returning total RAM and CPU values for the entire k8s node, not the pod. So when we run on a node with 64GB of RAM, we'd have to specify the minimum memory for each taskq worker replica pod to be at least ~3072 MiB to meet the 5% threshold, which seems excessive.
We see an MR to make system resource customization customizable here: #95. We'd be interested in picking that up so that we could configure the limits when running on k8s nodes or other VM infrastructures, if Sysinfo is reporting total resources of the host rather than the pods.
The text was updated successfully, but these errors were encountered:
We're running taskq workers on kubernetes pods, but have encountered a less than ideal situation with the
hasFreeSystemResources
check. It appears that the Sysinfo checks are returning total RAM and CPU values for the entire k8s node, not the pod. So when we run on a node with 64GB of RAM, we'd have to specify the minimum memory for each taskq worker replica pod to be at least ~3072 MiB to meet the 5% threshold, which seems excessive.We see an MR to make system resource customization customizable here: #95. We'd be interested in picking that up so that we could configure the limits when running on k8s nodes or other VM infrastructures, if Sysinfo is reporting total resources of the host rather than the pods.
The text was updated successfully, but these errors were encountered: