Allow setting metallb CIDR value for local env

[mikenairn]

The default CIDR value (/28) allows for 16 ips to be assigned which is fine for the majority of dev/test/demo needs. However, when trying to test at scale it is useful to be able to create more Gateways on a single cluster and therefore more IPs can be required.

Changes here allow the CIDR and number of IP values used by the local setup script to be set as needed, default is the same as before (/28 & 16 IPs).

make local-setup SUBNET_OFFSET=0 CIDR=26 NUM_IPS=64

[Boomatang]

If a users was to hit this issue, what would it look like? How would they identify the issue?

[mikenairn]

If a users was to hit this issue, what would it look like? How would they identify the issue?

If you created more gateways than IPs in the range, some Gateways will never go into a "Programmed" state:

kubectl get gateway -A
NAMESPACE        NAME                      CLASS   ADDRESS       PROGRAMMED   AGE
gateway-system   kuadrant-ingressgateway   istio   172.18.0.0    True         42m
scale-test-0     gw1-i0                    istio   172.18.0.2    True         31m
scale-test-0     gw2-i0                    istio   172.18.0.3    True         31m
scale-test-0     gw3-i0                    istio   172.18.0.4    True         31m
scale-test-0     gw4-i0                    istio   172.18.0.5    True         31m
scale-test-0     gw5-i0                    istio   172.18.0.6    True         31m
scale-test-1     gw1-i1                    istio   172.18.0.7    True         31m
scale-test-1     gw2-i1                    istio   172.18.0.8    True         31m
scale-test-1     gw3-i1                    istio   172.18.0.9    True         31m
scale-test-1     gw4-i1                    istio   172.18.0.10   True         31m
scale-test-1     gw5-i1                    istio   172.18.0.11   True         31m
scale-test-2     gw1-i2                    istio   172.18.0.12   True         31m
scale-test-2     gw2-i2                    istio   172.18.0.13   True         31m
scale-test-2     gw3-i2                    istio   172.18.0.14   True         31m
scale-test-2     gw4-i2                    istio   172.18.0.15   True         31m
scale-test-2     gw5-i2                    istio                 False        31m
scale-test-3     gw1-i3                    istio                 False        31m
scale-test-3     gw2-i3                    istio                 False        31m
scale-test-3     gw3-i3                    istio                 False        31m
scale-test-3     gw4-i3                    istio                 False        31m
scale-test-3     gw5-i3                    istio                 False        31m
scale-test-4     gw1-i4                    istio                 False        31m
scale-test-4     gw2-i4                    istio                 False        31m
scale-test-4     gw3-i4                    istio                 False        31m
scale-test-4     gw4-i4                    istio                 False        31m
scale-test-4     gw5-i4                    istio                 False        31m

Status will show that it's still pending:

kubectl get gateway/gw5-i2 -n scale-test-2 -o json | jq '.status.conditions[] | select(.type=="Programmed")'
{
  "lastTransitionTime": "2025-01-13T16:12:42Z",
  "message": "Assigned to service(s) gw5-i2-istio.scale-test-2.svc.cluster.local:80, but failed to assign to all requested addresses: address pending for hostname \"gw5-i2-istio.scale-test-2.svc.cluster.local\"",
  "observedGeneration": 1,
  "reason": "AddressNotAssigned",
  "status": "False",
  "type": "Programmed"
}

And if you want you can check the metallb logs:

kubectl logs deployment/controller -n metallb-system | grep "error"
{"caller":"service.go:140","error":"no available IPs","level":"error","msg":"IP allocation failed","op":"allocateIPs","ts":"2025-01-13T16:13:02Z"}