feat(netpol): authorize l3 protocols#5745
Conversation
|
I'll wait for #5741 to be merged to factorize the code I'll also add a global switch to enable this or to keep the default behaviour. And also an annotation on NetworkPolicies to act as a toggle. |
ab15e80 to
fdf56c4
Compare
Pull Request Test Coverage Report for Build 18551492763Warning: This coverage report may be inaccurate.This pull request's base commit is no longer the HEAD commit of its target branch. This means it includes changes from outside the original pull request, including, potentially, unrelated coverage changes.
Details
💛 - Coveralls |
fdf56c4 to
4b8b682
Compare
9900a58 to
b745109
Compare
Signed-off-by: SkalaNetworks <contact@skala.network>
b745109 to
dfe76cd
Compare
Signed-off-by: SkalaNetworks <contact@skala.network>
Signed-off-by: SkalaNetworks <contact@skala.network>
2e0b436 to
f3c0c57
Compare
Signed-off-by: SkalaNetworks <contact@skala.network>
|
@oilbeater new PR to address the discussion we had on Slack There's now a global controller switch and an annotation on network policies to configure the enforcement of network policies. When the enforcement is "standard", everything works just like it always has: only what is in the policy is allowed, the rest is blocked (including ICMP) When the enforcement is "lax", some protocols are excluded from being blocked. Those protocols are considered somewhat essential in a traditional network (especially one that has VMs in it): ARP/NDP (through ICMPv6)/ICMP/DHCP. That setting is especially useful when using VMs through Kubevirt, as they may rely on DHCP to get their MTU/IP/DNS and more from Kubevirt or Kube-OVN. ICMP is unnecessary in the standard Kubernetes networking model, but very important for VMs to do PMTU discovery. |
Signed-off-by: SkalaNetworks <contact@skala.network>
Pull Request
What type of this PR
Examples of user facing changes:
Which issue(s) this PR fixes
Fixes #(issue-number)