2018-04-24 09:29:30 UTC - Christos Anagnostakis: @Christos Anagnostakis has joined the channel ---- 2018-04-24 09:31:20 UTC - Byron: @jia zhai thanks. this was a fresh k8s deployment (albeit torn down and redeployed several times). @Sijie Guo using a daemonset.. the generic set: <https://github.com/apache/incubator-pulsar/tree/master/deployment/kubernetes/generic> ---- 2018-04-24 09:32:23 UTC - Byron: the pods were destroyed, so my assumption is that the local storage (not persistent) should have been wiped along with the pods ---- 2018-04-24 12:31:45 UTC - Byron: so it turns out that when using `hostPath` for a volume.. it will retain the data on the host even after the pod is deleted. it _makes sense_ in how this feature is described, but i hadn’t used this before. from the docs.
> A hostPath volume mounts a file or directory from the host node’s filesystem into your pod. This is not something that most Pods will need, but it offers a powerful escape hatch for some applications. For the provided k8s configs in the repo, it may be better to not include volumes by default to prevent this from happening to newcomers. In the docs, there would be a warning stating that the configs do not provide cross restart persistent and that needs to be added. ---- 2018-04-24 17:39:47 UTC - Sijie Guo: @Byron yeah, it should retain the data among restarts. good point about the warning. will file a github issue to improve it. back to your question, were you deploying a zookeeper cluster and delay the zookeeper cluster? I am suspecting this comes from mismatch between data and metadata. ---- 2018-04-25 08:53:35 UTC - Tao: @Tao has joined the channel ----
