fuel-ccp-installer/doc/source/troubleshooting.rst

Troubleshooting

Calico related problems

If you use standalone bare metal servers, or if you experience issues with a Calico bird daemon and networking for a Kubernetes cluster VMs, ensure that netfilter for bridge interfaces is disabled for your host node(s):

echo 0 > /proc/sys/net/bridge/bridge-nf-call-iptables

Otherwise, bird daemon inside Calico won't function correctly because of libvirt and NAT networks. More details can be found in this bug.

Then reporting issues, please also make sure to include details on the host OS type and its kernel version.

DNS resolve issues

See a known configuration issue. The workaround is as simple as described in the bug: always define custom intranet DNS resolvers in the upstream_dns_servers var listed in the first place, followed by public internet resolvers, if any.

Network check

While a net check is a part of deployment process, you can run the basic DNS check manually from a cluster node as bash /usr/local/bin/test_networking.sh. You can as well run all network checks from the admin node:

export ws=/home/workspace/
/usr/bin/ansible-playbook -e ansible_ssh_pass=vagrant -u vagrant -b \
--become-user=root -i ~${ws}inventory/inventory.cfg \
-e @${ws}kargo/inventory/group_vars/all.yml \
-e @${ws}inventory/kargo_default_common.yaml \
-e @${ws}inventory/kargo_default_ubuntu.yaml \
-e @${ws}inventory/custom.yaml \
${ws}utils/kargo/postinstall.yml -v --tags netcheck

There is also K8s netcheck server and agents applications running, if the cluster was created with the deploy_netchecker: true. In order to verify networking health and status of agents, which include timestamps of the last known healthy networking state, those may be quieried from any cluster node with:

curl -s -X GET 'http://localhost:31081/api/v1/agents/' | \
python -mjson.tool
curl -X GET 'http://localhost:31081/api/v1/connectivity_check'