Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion charts/shield/Chart.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -13,5 +13,5 @@ maintainers:
- name: mavimo
email: marcovito.moscaritolo@sysdig.com
type: application
version: 1.49.3
version: 1.50.0
appVersion: "1.0.0"
28 changes: 28 additions & 0 deletions charts/shield/tests/host/daemonset_test.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -1057,3 +1057,31 @@ tests:
mountPath: /opt/draios/etc/local_forwarder_config.yaml
subPath: local_forwarder_config.yaml
template: host/daemonset.yaml

- it: No CPU limit on the host shield by default
asserts:
- equal:
path: spec.template.spec.containers[?(@.name == "sysdig-host-shield")].resources
value:
limits:
memory: 1Gi
requests:
cpu: 250m
memory: 384Mi
template: host/daemonset.yaml
- notExists:
path: spec.template.spec.containers[?(@.name == "sysdig-host-shield")].resources.limits.cpu
template: host/daemonset.yaml

- it: CPU limit is applied when explicitly set
set:
host:
resources:
shield:
limits:
cpu: 4000m
asserts:
- equal:
path: spec.template.spec.containers[?(@.name == "sysdig-host-shield")].resources.limits.cpu
value: 4000m
template: host/daemonset.yaml
9 changes: 7 additions & 2 deletions charts/shield/values.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -360,9 +360,14 @@ host:
# The memory request for the kmodule
memory: 384Mi
shield:
# No CPU limit is set by default. The host shield sits in the syscall path and drains one
# ring buffer per CPU, so its work scales with the node's vCPU count and with container
# churn. A fixed CPU limit throttles the drain without reducing the per-syscall cost that
# is charged to the calling container, and on large or high-churn nodes the resulting
# event drops and container-runtime query backlog can degrade the node as a whole.
# Set a CPU limit only if your policy requires one, and size it to the node rather than
# to a flat value: `host.resources.shield.limits.cpu: 4000m`.
limits:
# The CPU limit for the host shield
cpu: 1000m
# The memory limit for the host shield
memory: 1Gi
requests:
Expand Down
Loading