Open sigram opened 4 years ago
I noticed that the default setting for docker-edge is to use osxfs, which results in high CPU usage. It would be nice to mention that users can switch to gRPC FUSE by checking the option "Include VM in Time Machine backups" in General > Settings.
Also, due to numerous bugs in com.docker.hyperkit when using the Kubernetes option in Docker.app there's significant CPU overhead, around 200% or more.
Thanks for finding these issues! It's easy to miss some of the setup steps when you are writing these instructions after-the-fact.
For posterity, @sigram and I have also been talking about using microk8s in the tutorial, which relates to #111. Some of these changes are applicable regardless, but the first two certainly would change with the shift to microk8s.
Additional thing I found going through the tutorial that I think fits here in terms of small corrections
/solr
in the Execute the Collection API command# Execute the Collections API command
curl "http://default-example-solrcloud.ing.local.domain/solr/admin/collections?action=CREATE&name=mycoll&numShards=1&replicationFactor=3&maxShardsPerNode=2&collection.configName=_default"
# Check in Admin UI that collection is created
open "http://default-example-solrcloud.ing.local.domain/solr/#/~cloud?view=graph"
Describe the bug Local tutorial on a fresh Mac doesn't work, it needs additional steps.
Environment (please complete the following information):
Additions & corrections to the tutorial
before calling
brew cask install docker-edge
the following was required, otherwise brew couldn't find this cask:brew tap homebrew/cask-versions
when you first run the Docker.app you need to check "Enable Kubernetes" in the Kubernetes tab, and "Apply & Restart".
"visiting the Kubernetes dashboard" - for a Kubernetes newbie this is totally cryptic, it should probably direct the user to the later section on installing the dashboard.
"Create a collection" section should probably add
?commit=true
when indexing new docs to make them immediately visible for searching. For Solr newbies it may be confusing that search will still return 0 docs for a while.following the example "scaling up" section, adding more (docker) replicas resulted in some instability - occasionally instead of 5 it would show 4 nodes, and the
kubectl proxy
would show:http: proxy error: context canceled