There was a bug in handling "default-allow" ipsets - it could not
cope with out of order receive of Pod updates. E.g:
1) AddPod(Pod_A, PodIP_A),
2) AddPod(Pod_B, PodIP_A),
3) DeletePod(Pod_A),
would result with PodIP_A being removed from "default-allow" ipset,
as such ipset entries were not refcounted. Bypassing of refcounting
was due to the fact that adding a netpol which dst-selects a Pod which
IP addr is in "default-allow" ipset has to remove the IP addr from
the ipset regardless of refcount value of the IP addr entry in the
ipset. So, refcounting in such case didn't make sense and it was
bypassed.
This commit changes the way entries are refcounted in ipsets, and
it makes "default-allow" ipset the same citizen as other ipsets, so
from now "default-allow" ipset is refcounted.
The main difference between new and legacy (pre-1.7) netpol semantics
are the following:
* The DenyDefault ingress isolation annotation is obsoleted.
* Traffic is allowed to a pod if there are no netpol selecting
the pod.
See https://kubernetes.io/docs/api-reference/v1.6/#networkpolicyspec-v1beta1-extensions
vs https://kubernetes.io/docs/api-reference/v1.7/#networkpolicyspec-v1-networking
for more details.
weave-npc can operate either in the legacy or the new mode (default). To
use the legacy, pass "--use-legacy-netpol" arg when starting the npc.
In non-legacy mode we create for each Namespace a default-allow ipset
which stores IP addrs of such Pods which are not selected by any
NetworkPolicy.The ipsets are dst's in rules of the WEAVE-NPC-DEFAULT chain.
This behaviour was breaking the use of pod selectors for the 'from'
part of a rule.
Move checking of local pods into namespace.go so we can maintain
LocalIpSet correctly
Martynas Pumputis