# Snapshot Test Day June 15th ### Focus on CLI install tool helm mode ### Connectivity test after upgrade from 1.13 to 1.14-snap ``` Test Report Test [to-fqdns]: ❌ to-fqdns/pod-to-world/http-to-one.one.one.one-0: cilium-test/client-6f6788d7cc-95vsp (10.0.0.201) -> one.one.one.one-http (one.one.one.one:80) ❌ to-fqdns/pod-to-world/http-to-one.one.one.one-1: cilium-test/client2-bc59f56d5-smp89 (10.0.0.222) -> one.one.one.one-http (one.one.one.one:80) connectivity test failed: 8 tests failed ``` Still getting odd failures for "world" connectivity tests. Probabably external service issues. ### Clustermesh external workloads Still having problems here.. looks like the specific external workload secret is not being populated in cli helm mode. ``` $ cilium clustermesh enable --service-type NodePort $ cilium clustermesh vm status Trying to get secret clustermesh-apiserver-external-workload-cert by deprecated name clustermesh-apiserver-external-workload-certs Error: Unable to determine status: unable to get external workload secret to access clustermesh service ``` Missing secret `clustermesh-apiserver-external-workload-cert` ### Setting helm values with cilium ubprade Retest enabling new promethus metrics from clustermesh. Now possible with cli helm mode. Here's what works. ``` $ cilium install --version v1.14.0-snapshot.3 $ cilium status --wait $ cilium clustermesh enable $ cilium status --wait $ kubectl -n kube-system describe deploy/clustermesh-apiserver $ cilium upgrade --version v1.14.0-snapshot.3 --helm-set "clustermesh.apiserver.metrics.enabled=true,clustermesh.apiserver.metrics.etcd.enabled=true" --reuse-values $ cilium status --wait $ kubectl -n kube-system describe deploy/clustermesh-apiserver ``` #### The bad 1. you MUST specify --version or cilium upgrade will choose default cilium version even if you just want to change helm settings. Which effectively downgrades from snapshot to 1.13 in my testing. Very unexpected behavior here and will cause problems. 2. must use --reuse-values flag to ensure your just changing settings explicitly or else it blows away the autodetect helm settings, and will cause problems for the casual user. If we start using cilium upgrade in guidance will need to make sure we consistently include this option. #### The good lets me enable/disable new clustermesh metrics after `cilium clustermesh enable`.