1KUBERNETES(1)(kubernetes) KUBERNETES(1)(kubernetes)
2Eric Paris Jan 2015
6
9 kubectl drain - Drain node in preparation for maintenance
10
14 kubectl drain [OPTIONS]
15
19 Drain node in preparation for maintenance.
20 The given node will be marked unschedulable to prevent new pods from
23 arriving. 'drain' evicts the pods if the APIServer supports http://ku‐
24 bernetes.io/docs/admin/disruptions/ . Otherwise, it will use normal
25 DELETE to delete the pods. The 'drain' evicts or deletes all pods ex‐
26 cept mirror pods (which cannot be deleted through the API server). If
27 there are DaemonSet-managed pods, drain will not proceed without --ig‐
28 nore-daemonsets, and regardless it will not delete any DaemonSet-man‐
29 aged pods, because those pods would be immediately replaced by the Dae‐
30 monSet controller, which ignores unschedulable markings. If there are
31 any pods that are neither mirror pods nor managed by ReplicationCon‐
32 troller, ReplicaSet, DaemonSet, StatefulSet or Job, then drain will not
33 delete any pods unless you use --force. --force will also allow dele‐
34 tion to proceed if the managing resource of one or more pods is miss‐
35 ing.
36 'drain' waits for graceful termination. You should not operate on the
39 machine until the command completes.
40 When you are ready to put the node back into service, use kubectl un‐
43 cordon, which will make the node schedulable again.
44 http://kubernetes.io/images/docs/kubectl_drain.svg
47
51 --delete-emptydir-data=false Continue even if there are pods using
52 emptyDir (local data that will be deleted when the node is drained).
53 --delete-local-data=false Continue even if there are pods using
56 emptyDir (local data that will be deleted when the node is drained).
57 --disable-eviction=false Force drain to use delete, even if evic‐
60 tion is supported. This will bypass checking PodDisruptionBudgets, use
61 with caution.
62 --dry-run="none" Must be "none", "server", or "client". If client
65 strategy, only print the object that would be sent, without sending it.
66 If server strategy, submit server-side request without persisting the
67 resource.
68 --force=false Continue even if there are pods not managed by a
71 ReplicationController, ReplicaSet, Job, DaemonSet or StatefulSet.
72 --grace-period=-1 Period of time in seconds given to each pod to
75 terminate gracefully. If negative, the default value specified in the
76 pod will be used.
77 --ignore-daemonsets=false Ignore DaemonSet-managed pods.
80 --pod-selector="" Label selector to filter pods on the node
83 -l, --selector="" Selector (label query) to filter on
86 --skip-wait-for-delete-timeout=0 If pod DeletionTimestamp older
89 than N seconds, skip waiting for the pod. Seconds must be greater than
90 0 to skip.
91 --timeout=0s The length of time to wait before giving up, zero
94 means infinite
95
99 --add-dir-header=false If true, adds the file directory to the
100 header of the log messages
101 --alsologtostderr=false log to standard error as well as files
104 --application-metrics-count-limit=100 Max number of application
107 metrics to store (per container)
108 --as="" Username to impersonate for the operation
111 --as-group=[] Group to impersonate for the operation, this flag
114 can be repeated to specify multiple groups.
115 --azure-container-registry-config="" Path to the file containing
118 Azure container registry configuration information.
119 --boot-id-file="/proc/sys/kernel/random/boot_id" Comma-separated
122 list of files to check for boot-id. Use the first one that exists.
123 --cache-dir="/builddir/.kube/cache" Default cache directory
126 --certificate-authority="" Path to a cert file for the certificate
129 authority
130 --client-certificate="" Path to a client certificate file for TLS
133 --client-key="" Path to a client key file for TLS
136 --cloud-provider-gce-l7lb-src-cidrs=130.211.0.0/22,35.191.0.0/16
139 CIDRs opened in GCE firewall for L7 LB traffic proxy health
140 checks
141 --cloud-provider-gce-lb-src-cidrs=130.211.0.0/22,209.85.152.0/22,209.85.204.0/22,35.191.0.0/16
144 CIDRs opened in GCE firewall for L4 LB traffic proxy health
145 checks
146 --cluster="" The name of the kubeconfig cluster to use
149 --container-hints="/etc/cadvisor/container_hints.json" location of
152 the container hints file
153 --containerd="/run/containerd/containerd.sock" containerd endpoint
156 --containerd-namespace="k8s.io" containerd namespace
159 --context="" The name of the kubeconfig context to use
162 --default-not-ready-toleration-seconds=300 Indicates the tolera‐
165 tionSeconds of the toleration for notReady:NoExecute that is added by
166 default to every pod that does not already have such a toleration.
167 --default-unreachable-toleration-seconds=300 Indicates the tolera‐
170 tionSeconds of the toleration for unreachable:NoExecute that is added
171 by default to every pod that does not already have such a toleration.
172 --disable-root-cgroup-stats=false Disable collecting root Cgroup
175 stats
176 --docker="unix:///var/run/docker.sock" docker endpoint
179 --docker-env-metadata-whitelist="" a comma-separated list of envi‐
182 ronment variable keys matched with specified prefix that needs to be
183 collected for docker containers
184 --docker-only=false Only report docker containers in addition to
187 root stats
188 --docker-root="/var/lib/docker" DEPRECATED: docker root is read
191 from docker info (this is a fallback, default: /var/lib/docker)
192 --docker-tls=false use TLS to connect to docker
195 --docker-tls-ca="ca.pem" path to trusted CA
198 --docker-tls-cert="cert.pem" path to client certificate
201 --docker-tls-key="key.pem" path to private key
204 --enable-load-reader=false Whether to enable cpu load reader
207 --event-storage-age-limit="default=0" Max length of time for which
210 to store events (per type). Value is a comma separated list of key val‐
211 ues, where the keys are event types (e.g.: creation, oom) or "default"
212 and the value is a duration. Default is applied to all non-specified
213 event types
214 --event-storage-event-limit="default=0" Max number of events to
217 store (per type). Value is a comma separated list of key values, where
218 the keys are event types (e.g.: creation, oom) or "default" and the
219 value is an integer. Default is applied to all non-specified event
220 types
221 --global-housekeeping-interval=1m0s Interval between global house‐
224 keepings
225 --housekeeping-interval=10s Interval between container housekeep‐
228 ings
229 --insecure-skip-tls-verify=false If true, the server's certificate
232 will not be checked for validity. This will make your HTTPS connections
233 insecure
234 --kubeconfig="" Path to the kubeconfig file to use for CLI re‐
237 quests.
238 --log-backtrace-at=:0 when logging hits line file:N, emit a stack
241 trace
242 --log-cadvisor-usage=false Whether to log the usage of the cAdvi‐
245 sor container
246 --log-dir="" If non-empty, write log files in this directory
249 --log-file="" If non-empty, use this log file
252 --log-file-max-size=1800 Defines the maximum size a log file can
255 grow to. Unit is megabytes. If the value is 0, the maximum file size is
256 unlimited.
257 --log-flush-frequency=5s Maximum number of seconds between log
260 flushes
261 --logtostderr=true log to standard error instead of files
264 --machine-id-file="/etc/machine-id,/var/lib/dbus/machine-id"
267 Comma-separated list of files to check for machine-id. Use the
268 first one that exists.
269 --match-server-version=false Require server version to match
272 client version
273 -n, --namespace="" If present, the namespace scope for this CLI
276 request
277 --one-output=false If true, only write logs to their native sever‐
280 ity level (vs also writing to each lower severity level
281 --password="" Password for basic authentication to the API server
284 --profile="none" Name of profile to capture. One of
287 (none|cpu|heap|goroutine|threadcreate|block|mutex)
288 --profile-output="profile.pprof" Name of the file to write the
291 profile to
292 --referenced-reset-interval=0 Reset interval for referenced bytes
295 (container_referenced_bytes metric), number of measurement cycles after
296 which referenced bytes are cleared, if set to 0 referenced bytes are
297 never cleared (default: 0)
298 --request-timeout="0" The length of time to wait before giving up
301 on a single server request. Non-zero values should contain a corre‐
302 sponding time unit (e.g. 1s, 2m, 3h). A value of zero means don't time‐
303 out requests.
304 -s, --server="" The address and port of the Kubernetes API server
307 --skip-headers=false If true, avoid header prefixes in the log
310 messages
311 --skip-log-headers=false If true, avoid headers when opening log
314 files
315 --stderrthreshold=2 logs at or above this threshold go to stderr
318 --storage-driver-buffer-duration=1m0s Writes in the storage driver
321 will be buffered for this duration, and committed to the non memory
322 backends as a single transaction
323 --storage-driver-db="cadvisor" database name
326 --storage-driver-host="localhost:8086" database host:port
329 --storage-driver-password="root" database password
332 --storage-driver-secure=false use secure connection with database
335 --storage-driver-table="stats" table name
338 --storage-driver-user="root" database username
341 --tls-server-name="" Server name to use for server certificate
344 validation. If it is not provided, the hostname used to contact the
345 server is used
346 --token="" Bearer token for authentication to the API server
349 --update-machine-info-interval=5m0s Interval between machine info
352 updates.
353 --user="" The name of the kubeconfig user to use
356 --username="" Username for basic authentication to the API server
359 -v, --v=0 number for the log level verbosity
362 --version=false Print version information and quit
365 --vmodule= comma-separated list of pattern=N settings for
368 file-filtered logging
369 --warnings-as-errors=false Treat warnings received from the server
372 as errors and exit with a non-zero exit code
373
377 # Drain node "foo", even if there are pods not managed by a ReplicationController, ReplicaSet, Job, DaemonSet or StatefulSet on it.
378 $ kubectl drain foo --force
379 # As above, but abort if there are pods not managed by a ReplicationController, ReplicaSet, Job, DaemonSet or StatefulSet, and use a grace period of 15 minutes.
381 $ kubectl drain foo --grace-period=900
382
387 kubectl(1),
388
392 January 2015, Originally compiled by Eric Paris (eparis at redhat dot
393 com) based on the kubernetes source material, but hopefully they have
394 been automatically generated since!
395Manuals User KUBERNETES(1)(kubernetes)
