Merge pull request kubernetes#387 from ddysher/cleanup-log

cleanup logs & update README
mqliang · Oct 13, 2016 · e967444 · e967444
2 parents afd98db + a69d611
commit e967444
Show file tree

Hide file tree

Showing 7 changed files with 39 additions and 44 deletions.
diff --git a/cluster/caicloud-anchnet/README.md b/cluster/caicloud-anchnet/README.md
@@ -104,21 +104,13 @@ used in kube-up. For a full list of options, consult the file.
 * `BUILD_VERSION`: The version to build during kube-up (for development mainly). Default value is current date and time, i.e.
   `$(TZ=Asia/Shanghai date +%Y-%m-%d-%H-%M)`.
 
-* `KUBE_INSTANCE_LOGDIR`: Directory for holding kubeup instance specific logs. During kube-up, instances will be installed/provisioned
-  concurrently; if we just send logs to stdout, stdout will mess up. Therefore, we specify a directory to hold instance specific logs.
-  All other logs will be sent to stdout. Default value is "`/tmp/kubeup-$(TZ=Asia/Shanghai date +%Y-%m-%d-%H-%M-%S)`". The log directory
-  looks like:
-  ```
-  $ ls /tmp/kubeup-2015-09-12-01-00-48
-  $ i-6TDSS52U i-ALCWC66X i-JZ9EDO70 i-THBJ0VCD i-UAAE4SUG
-  ```
-  where 'i-6TDSS52U' is instance id. Note the variable only catches logs for concurrent instance provisioning; all other logs, like
-  creating instances from anchnet, will be send to stdout. One common pattern is to set `KUBE_INSTANCE_LOGDIR` and redirect stdout to
-  `${KUBE_INSTANCE_LOGDIR}/common-logs`.
-
 * `ANCHNET_CONFIG_FILE`: The config file supplied to `anchnet` SDK CLI. Default value is `~/.anchnet/config`. Usually, you don't have
   to change the value, unless you want to create cluster under another anchnet account (NOT sub-account).
 
+* `KUBE_DISTRO`: The Linux distribution of underline OS, currently only ubuntu:trusty is supported.
+
+* `REGISTER_MASTER_KUBELET`: The flag controls whether we registry master as a node, this is true by default.
+
 * `ENABLE_CLUSTER_DNS`: Decide if cluster dns addon needs to be created, default to true. DNS addon is essential and should always be true.
 
 * `ENABLE_CLUSTER_LOGGING`: Decide if cluster logging addon needs to be created.
@@ -131,9 +123,8 @@ used in kube-up. For a full list of options, consult the file.
 
 ## Delete a cluster
 
-Deleting a cluster will delete everything associated with the cluster, including instances, vxnet, external ips, firewalls, etc. (TODO:
-delete loadbalancer, see [issue](https://github.com/caicloud/caicloud-kubernetes/issues/101)). To bring down a cluster, run:
-
+Deleting a cluster will delete everything associated with the cluster, including instances, vxnet, external ips, firewalls, etc. To bring
+down a cluster, run:
 ```
 KUBERNETES_PROVIDER=caicloud-anchnet ./cluster/kube-down.sh
 ```
@@ -163,21 +154,24 @@ KUBERNETES_PROVIDER=caicloud-anchnet ./cluster/kube-push.sh
 
 * `CAICLOUD_KUBE_VERSION`: Which version to deploy.
 
-
 ## Add node to a cluster
 
 In case we want to add node(s) to a running cluster, we can run:
 ```
 CLUSTER_NAME=kube-default MASTER_EIP=103.21.116.147 NUM_MINIONS=1 KUBERNETES_PROVIDER=caicloud-anchnet ./cluster/kube-add-node.sh
 ```
 
+Note this is not working as of now.
+
 ## Stop a running cluster
+
 we can simply do the following to shut down a running cluster:
 ```
 CLUSTER_NAME=kube-default KUBERNETES_PROVIDER=caicloud-anchnet ./cluster/kube-halt.sh
 ```
 
 ## Restart a cluster
+
 If we want to restart a cluster which has been stopped before, we can do:
 ```
 CLUSTER_NAME=kube-default KUBERNETES_PROVIDER=caicloud-anchnet ./cluster/kube-restart.sh

diff --git a/cluster/caicloud-anchnet/config-default.sh b/cluster/caicloud-anchnet/config-default.sh
@@ -103,12 +103,6 @@ REGISTER_MASTER_KUBELET=${REGISTER_MASTER_KUBELET:-"false"}
 # Following params in the section should rarely change.
 #
 
-# Directory for holding kubeup instance specific logs. During kube-up, instances
-# will be installed/provisioned concurrently; if we just send logs to stdout,
-# stdout will mess up. Therefore, we specify a directory to hold instance specific
-# logs. All other logs will be sent to stdout, e.g. create instances from anchnet.
-KUBE_INSTANCE_LOGDIR=${KUBE_INSTANCE_LOGDIR:-"/tmp/caicloud-kube-`TZ=Asia/Shanghai date +%Y-%m-%d-%H-%M-%S`"}
-
 # URL path of the server hosting caicloud kubernetes release.
 CAICLOUD_HOST_URL=${CAICLOUD_HOST_URL:-"http://7xli2p.dl1.z0.glb.clouddn.com"}
 

diff --git a/cluster/caicloud-anchnet/util.sh b/cluster/caicloud-anchnet/util.sh
@@ -101,10 +101,8 @@ function kube-up {
   # Make sure we have:
   #  1. a staging area
   #  2. ssh capability
-  #  3. log directory
   ensure-temp-dir
   ensure-ssh-agent
-  ensure-log-dir
 
   if [[ "${KUBE_UP_MODE}" = "dev" ]]; then
     # For dev, set to existing instance IDs for master and nodes. Other variables
@@ -1039,7 +1037,6 @@ function create-anchnet-config {
 #   NODE_INSTANCE_IDS_ARR
 #   NODE_IIPS_ARR
 #   PRIVATE_SDN_INTERFACE
-#   KUBE_INSTANCE_LOGDIR
 function setup-anchnet-instances {
   # Use multiple retries since seting up sdn network is essential for follow-up
   # installations, and we ses occational errors:

diff --git a/cluster/caicloud-baremetal/README.md b/cluster/caicloud-baremetal/README.md
@@ -0,0 +1,28 @@
+# Caicloud baremetal cloudprovider
+
+Baremetal cloudprovider implements kubernetes cloudprovider plugin. It assumes nothing but a few running instances.
+
+## Create a development cluster
+
+To create a kubernetes cluster for local developement, first create VMs using Vagrant file:
+```
+vagrant up
+```
+
+This will create two ubuntu VMs with dedicated IP addresses. Now, to bring up kubernete cluster, simply run:
+```
+KUBERNETES_PROVIDER=caicloud-baremetal ./cluster/kube-up.sh
+```
+
+#### Options:
+
+There are quite a few options used to create a baremetal cluster, located at `config-default.sh`. Following is a curated list of options
+used in kube-up. For a full list of options, consult the file.
+
+* `KUBE_DISTRO`: The Linux distribution of underline OS, currently only ubuntu:trusty is supported.
+
+* `MASTER_SSH_INFO`: The master ssh information in the format of "username:password@ip_address". HA master is currently supported.
+
+* `NODE_SSH_INFO`: The master ssh information in the format of "username:password@ip_address". Multiple nodes are separated via comma.
+
+* `REGISTER_MASTER_KUBELET`: The flag controls whether we registry master as a node, this is true by default.
diff --git a/cluster/caicloud-baremetal/config-default.sh b/cluster/caicloud-baremetal/config-default.sh
@@ -20,7 +20,7 @@
 # Linux distribution of underline machines.
 KUBE_DISTRO=${KUBE_DISTRO:-"trusty"}
 
-# ssh information for master.
+# ssh information for master (comma separated string).
 MASTER_SSH_INFO=${MASTER_SSH_INFO:-"vagrant:vagrant@192.168.205.10"}
 
 # ssh information for nodes (comma separated string).
@@ -70,12 +70,6 @@ REGISTER_MASTER_KUBELET=${REGISTER_MASTER_KUBELET:-"true"}
 # Following params in the section should rarely change.
 #
 
-# Directory for holding kubeup instance specific logs. During kube-up, instances
-# will be installed/provisioned concurrently; if we just send logs to stdout,
-# stdout will mess up. Therefore, we specify a directory to hold instance specific
-# logs. All other logs will be sent to stdout, e.g. create instances from anchnet.
-KUBE_INSTANCE_LOGDIR=${KUBE_INSTANCE_LOGDIR:-"/tmp/kubeup-`TZ=Asia/Shanghai date +%Y-%m-%d-%H-%M-%S`"}
-
 # URL path of the server hosting caicloud kubernetes release.
 CAICLOUD_HOST_URL=${CAICLOUD_HOST_URL:-"http://7xli2p.dl1.z0.glb.clouddn.com"}
 

diff --git a/cluster/caicloud/common.sh b/cluster/caicloud/common.sh
@@ -554,7 +554,6 @@ sudo chmod go-rwx /etc/caicloud"
 # to master without using password.
 #
 # Assumed vars:
-#   KUBE_INSTANCE_LOGDIR
 #   CAICLOUD_TARBALL_URL
 #   MASTER_SSH_EXTERNAL *
 #   NODE_SSH_EXTERNAL *
@@ -834,16 +833,6 @@ function ensure-ssh-agent {
   fi
 }
 
-# Make sure log directory exists.
-#
-# Assumed vars
-#   KUBE_INSTANCE_LOGDIR
-function ensure-log-dir {
-  if [[ ! -z ${KUBE_INSTANCE_LOGDIR-} ]]; then
-    mkdir -p ${KUBE_INSTANCE_LOGDIR}
-  fi
-}
-
 # Make sure ~/kube exists on the master/node. This is used in kube-push
 # because we now clean up ~/kube directory once the cluster is up and running.
 #

diff --git a/cluster/caicloud/trusty/helper.sh b/cluster/caicloud/trusty/helper.sh
@@ -357,7 +357,6 @@ EOF
 # Assumed vars:
 #   INSTANCE_SSH_EXTERNAL
 #   APT_MIRRORS
-#   KUBE_INSTANCE_LOGDIR
 function install-packages {
   APT_MIRROR_INDEX=0            # Used for choosing an apt mirror.
   command-exec-and-retry "install-packages-internal" 2 "${1:-}"