Skip to content
Projects
Groups
Snippets
Help
This project
Loading...
Sign in / Register
Toggle navigation
K
k3s
Project
Project
Details
Activity
Cycle Analytics
Repository
Repository
Files
Commits
Branches
Tags
Contributors
Graph
Compare
Charts
Issues
0
Issues
0
List
Board
Labels
Milestones
Merge Requests
0
Merge Requests
0
CI / CD
CI / CD
Pipelines
Jobs
Schedules
Charts
Registry
Registry
Wiki
Wiki
Snippets
Snippets
Members
Members
Collapse sidebar
Close sidebar
Activity
Graph
Charts
Create a new issue
Jobs
Commits
Issue Boards
Open sidebar
Jacklull
k3s
Commits
63dece9b
Commit
63dece9b
authored
Apr 21, 2015
by
Robert Bailey
Browse files
Options
Browse Files
Download
Plain Diff
Merge pull request #7135 from GoogleCloudPlatform/revert-7124-revert-validate.fix
Revert "Rollback "validate-cluster.sh" to previous version."
parents
e0b754d5
8745ebce
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
31 additions
and
61 deletions
+31
-61
validate-cluster.sh
cluster/validate-cluster.sh
+31
-61
No files found.
cluster/validate-cluster.sh
View file @
63dece9b
...
@@ -14,11 +14,7 @@
...
@@ -14,11 +14,7 @@
# See the License for the specific language governing permissions and
# See the License for the specific language governing permissions and
# limitations under the License.
# limitations under the License.
# Bring up a Kubernetes cluster.
# Validates that the cluster is healthy.
#
# If the full release name (gs://<bucket>/<release>) is passed in then we take
# that directly. If not then we assume we are doing development stuff and take
# the defaults in the release config.
set
-o
errexit
set
-o
errexit
set
-o
nounset
set
-o
nounset
...
@@ -28,12 +24,9 @@ KUBE_ROOT=$(dirname "${BASH_SOURCE}")/..
...
@@ -28,12 +24,9 @@ KUBE_ROOT=$(dirname "${BASH_SOURCE}")/..
source
"
${
KUBE_ROOT
}
/cluster/kube-env.sh"
source
"
${
KUBE_ROOT
}
/cluster/kube-env.sh"
source
"
${
KUBE_ROOT
}
/cluster/
${
KUBERNETES_PROVIDER
}
/util.sh"
source
"
${
KUBE_ROOT
}
/cluster/
${
KUBERNETES_PROVIDER
}
/util.sh"
get-password
detect-master
>
/dev/null
detect-minions
>
/dev/null
MINIONS_FILE
=
/tmp/minions-
$$
MINIONS_FILE
=
/tmp/minions-
$$
trap
'rm -rf "${MINIONS_FILE}"'
EXIT
trap
'rm -rf "${MINIONS_FILE}"'
EXIT
# Make several attempts to deal with slow cluster birth.
# Make several attempts to deal with slow cluster birth.
attempt
=
0
attempt
=
0
while
true
;
do
while
true
;
do
...
@@ -54,62 +47,39 @@ done
...
@@ -54,62 +47,39 @@ done
echo
"Found
${
found
}
nodes."
echo
"Found
${
found
}
nodes."
cat
-n
"
${
MINIONS_FILE
}
"
cat
-n
"
${
MINIONS_FILE
}
"
# On vSphere, use minion IPs as their names
attempt
=
0
if
[[
"
${
KUBERNETES_PROVIDER
}
"
==
"vsphere"
||
"
${
KUBERNETES_PROVIDER
}
"
==
"vagrant"
||
"
${
KUBERNETES_PROVIDER
}
"
==
"libvirt-coreos"
||
"
${
KUBERNETES_PROVIDER
}
"
==
"juju"
]]
;
then
while
true
;
do
MINION_NAMES
=(
"
${
KUBE_MINION_IP_ADDRESSES
[@]
}
"
)
kubectl_output
=
$(
"
${
KUBE_ROOT
}
/cluster/kubectl.sh"
get cs
)
fi
# On AWS we can't really name the minions, so just trust that if the number is right, the right names are there.
if
[[
"
${
KUBERNETES_PROVIDER
}
"
==
"aws"
]]
;
then
MINION_NAMES
=(
"
$(
cat
${
MINIONS_FILE
})
"
)
# /healthz validation isn't working for some reason on AWS. So just hope for the best.
# TODO: figure out why and fix, it must be working in some form, or else clusters wouldn't work.
echo
"Kubelet health checking on AWS isn't currently supported, assuming everything is good..."
echo
-e
"
${
color_green
}
Cluster validation succeeded
${
color_norm
}
"
exit
0
fi
for
((
i
=
0
;
i<
${#
MINION_NAMES
[@]
}
;
i++
))
;
do
# The "kubectl componentstatuses" output is four columns like this:
# Grep returns an exit status of 1 when line is not found, so we need the : to always return a 0 exit status
#
count
=
$(
grep
-c
"
${
MINION_NAMES
[
$i
]
}
"
"
${
MINIONS_FILE
}
"
)
||
:
# COMPONENT HEALTH MSG ERR
if
[[
"
${
count
}
"
==
"0"
]]
;
then
# controller-manager Healthy ok nil
echo
-e
"
${
color_red
}
Failed to find
${
MINION_NAMES
[
$i
]
}
, cluster is probably broken.
${
color_norm
}
"
#
cat
-n
"
${
MINIONS_FILE
}
"
# Parse the output to capture the value of the second column("HEALTH"), then use grep to
exit
1
# count the number of times it doesn't match "success".
fi
# Because of the header, the actual unsuccessful count is 1 minus the count.
name
=
"
${
MINION_NAMES
[
$i
]
}
"
non_success_count
=
$(
echo
"
${
kubectl_output
}
"
|
\
if
[[
"
$KUBERNETES_PROVIDER
"
!=
"vsphere"
&&
"
$KUBERNETES_PROVIDER
"
!=
"vagrant"
&&
"
$KUBERNETES_PROVIDER
"
!=
"libvirt-coreos"
&&
"
$KUBERNETES_PROVIDER
"
!=
"juju"
]]
;
then
sed
-n
's/^\([[:alnum:][:punct:]]\+\)\s\+\([[:alnum:][:punct:]]\+\)\s\+.*/\2/p'
|
\
# Grab fully qualified name
grep
'Healthy'
--invert-match
-c
)
name
=
$(
grep
"
${
MINION_NAMES
[
$i
]
}
\.
"
"
${
MINIONS_FILE
}
"
)
fi
# Make sure the kubelet is healthy.
if
((
non_success_count
>
1
))
;
then
# Make several attempts to deal with slow cluster birth.
if
((
attempt < 5
))
;
then
attempt
=
0
echo
-e
"
${
color_yellow
}
Cluster not working yet.
${
color_norm
}
"
while
true
;
do
echo
-n
"Attempt
$((
attempt+1
))
at checking Kubelet installation on node
${
MINION_NAMES
[
$i
]
}
..."
if
[[
"
$KUBERNETES_PROVIDER
"
!=
"libvirt-coreos"
&&
"
$KUBERNETES_PROVIDER
"
!=
"juju"
]]
;
then
curl_output
=
$(
curl
-s
--insecure
--user
"
${
KUBE_USER
}
:
${
KUBE_PASSWORD
}
"
\
"https://
${
KUBE_MASTER_IP
}
/api/v1beta1/proxy/minions/
${
name
}
/healthz"
)
else
curl_output
=
$(
curl
-s
\
"http://
${
KUBE_MASTER_IP
}
:8080/api/v1beta1/proxy/minions/
${
name
}
/healthz"
)
fi
if
[[
"
${
curl_output
}
"
!=
"ok"
]]
;
then
if
((
attempt
>
5
))
;
then
echo
echo
-e
"
${
color_red
}
Kubelet failed to install on node
${
MINION_NAMES
[
$i
]
}
. Your cluster is unlikely to work correctly."
echo
-e
"Please run ./cluster/kube-down.sh and re-create the cluster. (sorry!)
${
color_norm
}
"
exit
1
fi
else
echo
-e
"
${
color_green
}
[working]
${
color_norm
}
"
break
fi
echo
-e
"
${
color_yellow
}
[not working yet]
${
color_norm
}
"
attempt
=
$((
attempt+1
))
attempt
=
$((
attempt+1
))
sleep
30
sleep
30
done
else
echo
-e
"
${
color_yellow
}
Validate output:
${
color_norm
}
"
echo
"
${
kubectl_output
}
"
echo
-e
"
${
color_red
}
Validation returned one or more failed components. Cluster is probably broken.
${
color_norm
}
"
exit
1
fi
else
break
fi
done
done
echo
"Validate output:"
echo
"
${
kubectl_output
}
"
echo
-e
"
${
color_green
}
Cluster validation succeeded
${
color_norm
}
"
echo
-e
"
${
color_green
}
Cluster validation succeeded
${
color_norm
}
"
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment