Commit Graph

1095 Commits (51e4ccf10657e0649f9c038e2621915b6ccd647d)

Author SHA1 Message Date
k8s-merge-robot d42030170b Merge pull request #15999 from aveshagarwal/master-issue-15624
Auto commit by PR queue bot
2015-11-24 17:07:10 -08:00
Wojciech Tyczynski b6ef62af24 Use unversioned.ListOptions in clients. 2015-11-24 16:52:09 +01:00
Avesh Agarwal 3d5207fd73 This commit adds type information to events.
This addresses issue #15624.
2015-11-24 09:59:54 -05:00
k8s-merge-robot 683e908f0a Merge pull request #17265 from gambol99/rj/node_labels
Auto commit by PR queue bot
2015-11-24 06:55:31 -08:00
Yifan Gu 28e78f4fb8 rkt: remove rkt image manager.
This removes the dead code of rkt image manager as the generic
image manager is being used.
2015-11-20 16:30:51 -08:00
Yu-Ju Hong 41a9ab388d Use 10s period for image pulling backoff
This is consistent with the container restart backoff period.
2015-11-20 11:57:56 -08:00
Rohith c2526c93fa - adding the -node-label flag to the kubelet which allows for a initial tagging / labelling of the node on cluster registration
- the labels can come from a series of key=pair value or file:///path_to_file which contains key pairs
2015-11-18 23:20:58 +00:00
k8s-merge-robot 9604988537 Merge pull request #16894 from alexhersh/manifest-fix
Auto commit by PR queue bot
2015-11-17 07:39:15 -08:00
k8s-merge-robot 1334401888 Merge pull request #17202 from brendandburns/fix2
Auto commit by PR queue bot
2015-11-17 00:49:13 -08:00
cephaslr aef7af1ba4 Rename kubelet variable pods to maxPods in kubelet pkg to match pods variable in kubelet cmd.
Remove comment requesting modification of mesos contrib file, dependencies should be maintained in unit tests
2015-11-16 20:15:40 -05:00
Jeff Lowdermilk 70d89a3541 Merge pull request #13571 from yujuhong/lifecycle_v0
kubelet: add a generic pod lifecycle event generator
2015-11-16 14:59:40 -08:00
k8s-merge-robot 7fa85d430b Merge pull request #16614 from pmorie/ro-managed
Auto commit by PR queue bot
2015-11-16 10:46:44 -08:00
k8s-merge-robot 081b21687e Merge pull request #16261 from zhengguoyong/update_condition
Auto commit by PR queue bot
2015-11-16 02:34:18 -08:00
Paul Morie 911757486d Refactor volume.Builder.IsReadOnly() to volume.Builder.GetAttributes() 2015-11-13 23:34:48 -05:00
Brendan Burns fb576f30c8 Refactor an interface for style 2015-11-13 15:56:27 -08:00
Yu-Ju Hong ac778e8203 Adjust the sync/backoff period
Set resyncInterval to one minute now that we rely on the generic pleg to trigger
pod syncs on container events. When there is an error during syncing, pod
workers need to wake up sooner to retry. Set the sync error backoff period to
10 second in this case.
2015-11-13 09:57:25 -08:00
Yu-Ju Hong bc6414a873 kubelet: add a generic pod lifecycle event generator
This change introduces pod lifecycle event generator (PLEG), and adds a generic
PLEG. The generic PLEG relies on relisting to discover container events, and is
container-runtime-agnostic. Both docker and rkt are changed to use generic
PLEG.
2015-11-13 09:55:36 -08:00
Alexander Hersh 0584f9ba7a Create mirrorPod in runOnce to update API before syncPod
+ Fix #14992
+ "When deploying a pod using an on-disk kubelet manifest (a la /etc/kubernetes/manifests), it appears that the network plugin setUpPod is notified of the new pod before the apiserver."
2015-11-12 15:35:45 -08:00
k8s-merge-robot e88593d788 Merge pull request #14542 from vishh/kubelet-refactor
Auto commit by PR queue bot
2015-11-12 15:24:22 -08:00
k8s-merge-robot fbbc5a85a9 Merge pull request #16782 from Random-Liu/past-active-deadline-check
Auto commit by PR queue bot
2015-11-12 14:01:45 -08:00
k8s-merge-robot e80dd6689f Merge pull request #16866 from justinsb/zone_labels_on_nodes
Auto commit by PR queue bot
2015-11-12 07:25:37 -08:00
Vishnu kannan b1770537ab Make cadvisor startup synchronous with container runtime initialization.
This is hopefully a temporary workaround.

Signed-off-by: Vishnu kannan <vishnuk@google.com>
2015-11-11 15:20:13 -08:00
Vishnu kannan 4ad3d6f5fe Move container manager into a separate package.
Inject container manager into Kubelet. This lets us stub out container
manager during integration testing.
2015-11-11 15:00:37 -08:00
Vishnu kannan 129dbc734c avoid requiring runtime to be up while initializing kubelet image manager 2015-11-11 15:00:37 -08:00
Vishnu kannan 89c3cb2f43 refactor podCIDR handling in kubelet to runtime state 2015-11-11 15:00:37 -08:00
Vishnu kannan 5f4570b764 handling locks and update tests. Fail node if network is not initialized 2015-11-11 15:00:37 -08:00
Vishnu kannan cf56f7a8ef Refactoring kubelet to separate object creation from object logical initialization. 2015-11-11 15:00:34 -08:00
k8s-merge-robot 8761ad3ec1 Merge pull request #15464 from jiangyaoguo/bubble-up-reason-when-killing-pod
Auto commit by PR queue bot
2015-11-11 14:17:27 -08:00
Lantao Liu d6b93cdfe1 Move active deadline check into main syncLoop of kubelet. 2015-11-11 10:05:04 -08:00
k8s-merge-robot 0ee52eb0c8 Merge pull request #17102 from zhengguoyong/add_break
Auto commit by PR queue bot
2015-11-11 03:57:16 -08:00
zhengguoyong 79d97abac4 When find NodeReadyCondition, it can quit the loop 2015-11-11 16:15:01 +08:00
Tim St. Clair 67cfed5bf3 Don't wait for sync to update readiness
Push status updates as soon as readiness state changes for containers,
rather than waiting for the sync loop to update the status. In
particular, this should help new containers to come online faster.

Additionally, consolidates prober test helpers into a single file.
2015-11-10 14:00:12 -08:00
eulerzgy 3f630d8ad7 when pod has successed, update condition to PodCompleted 2015-11-10 15:37:44 +08:00
jiangyaoguo b0f0c294d9 make reasons constants with clarity 2015-11-09 23:42:14 +08:00
Wojciech Tyczynski 8a8f394f0d Merge pull request #16945 from kargakis/kubelet-fallback-to-apiserver
kubelet: Fallback to api server for pod status
2015-11-09 16:12:07 +01:00
k8s-merge-robot 5fe3733f95 Merge pull request #16272 from jiangyaoguo/remove-infra-container-event
Auto commit by PR queue bot
2015-11-09 06:05:35 -08:00
kargakis d519741e1b kubelet: Fallback to api server for pod status
This commit fixes getting the logs from complete/failed pods after
a kubelet restart by falling back to the api server in case we fail
to resolve the pod status using the status cache.
2015-11-06 21:15:40 +01:00
Tim St. Clair 1e88a682da Add liveness/readiness probe parameters
- PeriodSeconds - How often to probe
- SuccessThreshold - Number of successful probes to go from failure to success state
- FailureThreshold - Number of failing probes to go from success to failure state

This commit includes to changes in behavior:

1. InitialDelaySeconds now defaults to 10 seconds, rather than the
kubelet sync interval (although that also defaults to 10 seconds).
2. Prober only retries on probe error, not failure. To compensate, the
default FailureThreshold is set to the maxRetries, 3.
2015-11-06 10:46:40 -08:00
Justin Santa Barbara b2c2d617cf Label nodes with Zone information, if available
This lays the groundwork for simple multizone capabilities.

In a cloud environment, nodes are typically created by the kubelet
registering with the API server.  When creating a new node, we now query
the cloudprovider to see if it can provide Zone information, and if so
we add some well-known labels to the Node we are creating.
2015-11-05 14:35:05 -05:00
Yu-Ju Hong 5863f2a0a5 kubelet: reduce logging frequency for sync loop
Only logs when there are pods to sync.
2015-11-05 10:06:10 -08:00
k8s-merge-robot fb571e3e2e Merge pull request #16545 from timstclair/mirrorpods
Auto commit by PR queue bot
2015-11-05 02:56:23 -08:00
Tim St. Clair 858126b42a Clean up static/mirror pod status logic
- status.Manager always deals with the local (static) pod, but gets the
  mirror pod when syncing
  - This lets components like the probe workers ignore mirror pods
2015-11-04 11:42:25 -08:00
Yu-Ju Hong b734155954 Clean up housekeeping routine in kubelet
Now that kubelet checks sources seen correctly, there is no need to enforce the
initial order of pod updates and housekeeping. Use a ticker for housekeeping to
simplify the code.
2015-11-04 08:12:18 -08:00
jiangyaoguo 3ce5a51007 remove implicitly required container event 2015-11-04 16:47:17 +08:00
Yu-Ju Hong 2eb17df46b kubelet: independent pod syncs and backoff on error
Currently kubelet syncs all pods every 10s. This is not preferred because
 * Some pods may have been sync'd recently.
 * This may cause all the pods to be sync'd at once, causing undesirable
   CPU spikes.

This PR replaces the global syncs with independent, periodic pod syncs. At the
end of syncing, each pod worker will enqueue itslef with a future timestamp (
current time + sync interval), when it will be due for another sync.
 * If the pod worker encoutners an sync error, it may requeue with a different
   timestamp to retry sooner.
 * If a sync is triggered by the update channel (events or spec changes), the
   pod worker would enqueue a new sync time.

This change is necessary for moving to long or no periodic sync period once pod
lifecycle event generator is completed. We will still rely on the mechanism to
requeue the pod on sync error.

This change also makes sure that if a sync does not succeed (either due to
real error or the per-container backoff mechanism), an error would be propagated
back to the pod worker, which is responsible for requeuing.
2015-11-03 13:29:08 -08:00
mqliang 3afcbae1a8 remove unnecessary fmt 2015-11-02 21:07:07 +08:00
Dawn Chen a39e1e96dc Mark NodeNotReady for docker version below 1.6.2 2015-10-30 13:48:51 -07:00
Sami Wagiaalla 1d352a16b8 Support volume relabling for pods which specify an SELinux label 2015-10-28 09:26:58 -04:00
Daniel Smith 8524d392bf Merge pull request #16178 from madhusudancs/kubelet-out-of-disk-no-master
Report node out of disk condition in the kubelet.
2015-10-27 14:28:03 -07:00
k8s-merge-robot 0a7a9f3535 Merge pull request #16340 from timstclair/flaky-prober
Auto commit by PR queue bot
2015-10-27 12:10:22 -07:00
Jerzy Szczepkowski 5ef087adab Merge pull request #16266 from wojtek-t/fix_handshake_reflector
Reuse TCP connections in Reflector between resync periods.
2015-10-27 10:36:18 +01:00
Tim St. Clair 17b0a964f4 Always resync after resyncInterval 2015-10-26 18:50:57 -07:00
Madhusudan.C.S 9c4424f0bd Report out of disk as a node condition when node goes out of disk.
Define a new out of disk node condition and use it to report when node
goes out of disk.

Make a copy of loop range clause variable in node listers so that it
is available outside the for loop.

Also update/implement unit tests.
2015-10-26 15:01:20 -07:00
Daniel Smith e42f5af6a1 Merge pull request #16219 from thockin/resolvconf-filter
Add a cloud-provider hook to scrub DNS for pods
2015-10-26 12:32:02 -07:00
Wojciech Tyczynski d47e21f19f Reuse TCP connections in Reflector between resync periods. 2015-10-26 19:35:25 +01:00
k8s-merge-robot bffdd242d8 Merge pull request #16191 from yujuhong/fix_logs
Auto commit by PR queue bot
2015-10-26 09:31:35 -07:00
k8s-merge-robot 0e1ee14cc8 Merge pull request #15705 from wojtek-t/watch_options_in_client
Auto commit by PR queue bot
2015-10-24 17:39:20 -07:00
Abhi Shah 976cc12c8e Merge pull request #16174 from ArtfulCoder/nilCheck
Added Nil Check
2015-10-24 14:16:28 -07:00
k8s-merge-robot 5280540384 Merge pull request #15965 from Random-Liu/use-docker-label
Auto commit by PR queue bot
2015-10-24 13:19:40 -07:00
Wojciech Tyczynski f4d75e0a0a Support timeout in watch requests 2015-10-24 13:12:49 +02:00
k8s-merge-robot e1559e9fb5 Merge pull request #15352 from pmorie/fsgroup
Auto commit by PR queue bot
2015-10-24 01:55:45 -07:00
Tim Hockin 42c7fec490 Add a cloud-provider hook to scrub DNS for pods
GCE needs this hook and it seems general enough to include.
2015-10-23 17:01:49 -07:00
Yu-Ju Hong e4f739115f kubelet: fix getting logs from static pods
This change adds the necessary UID translation logic to fix the issue.
2015-10-23 14:31:40 -07:00
Lantao Liu 67a7633202 Put restart count into docker label 2015-10-23 13:02:32 -07:00
Abhishek Shah 4a7b4f2ed2 Nil Check 2015-10-23 11:52:59 -07:00
Filip Grzadkowski 165169ab1c Merge pull request #16052 from ArtfulCoder/mount_etc_hosts
kubelet manages /etc/hosts file
2015-10-23 14:07:34 +02:00
Filip Grzadkowski edd10d8a83 Merge pull request #15914 from vishh/serialize-pull
make kubelet image pulls serialized by default.
2015-10-23 14:04:23 +02:00
Vishnu kannan 0df4b46d4c Adding a kubelet flag to optionally enable parallel image pulls. 2015-10-22 17:19:51 -07:00
Paul Morie 3cd12f5e05 FSGroup implementation 2015-10-22 16:40:59 -04:00
Abhishek Shah ba6469d478 kubelet manages /etc/hosts file 2015-10-22 13:27:05 -07:00
k8s-merge-robot 6dc3dcf36a Merge pull request #15414 from thockin/exp-beta-annotations
Auto commit by PR queue bot
2015-10-20 17:45:32 -07:00
Tim St. Clair a263c77b65 Refactor liveness probing
This commit builds on previous work and creates an independent
worker for every liveness probe. Liveness probes behave largely the same
as readiness probes, so much of the code is shared by introducing a
probeType paramater to distinguish the type when it matters. The
circular dependency between the runtime and the prober is broken by
exposing a shared liveness ResultsManager, owned by the
kubelet. Finally, an Updates channel is introduced to the ResultsManager
so the kubelet can react to unhealthy containers immediately.
2015-10-19 15:15:59 -07:00
k8s-merge-robot 75c977d200 Merge pull request #15596 from zhengguoyong/alias_util_errors_packagename
Auto commit by PR queue bot
2015-10-19 02:35:37 -07:00
k8s-merge-robot 09a3871506 Merge pull request #15755 from zhengguoyong/del_capatical_packagename_for_cadvisorApi2
Auto commit by PR queue bot
2015-10-18 04:04:11 -07:00
eulerzgy f8f9afb874 alias local packagename for pkg/util/errors 2015-10-18 09:37:46 +08:00
k8s-merge-robot d471318aa8 Merge pull request #15799 from dchen1107/podstatus
Auto commit by PR queue bot
2015-10-16 19:32:30 -07:00
Dawn Chen 1b8f851979 Fix PodPhase issue caused by backoff 2015-10-16 15:16:07 -07:00
k8s-merge-robot 535090312f Merge pull request #13180 from samsabed/backoff_e2e
Auto commit by PR queue bot
2015-10-16 13:33:36 -07:00
eulerzgy 31c09bdcb8 Del capatical local packagename for cadvisorApi 2015-10-16 11:03:50 +08:00
Sam Abed 78979c77ce backoff e2e tests
Signed-off-by: Sam Abed <samsabed@gmail.com>
2015-10-16 13:09:29 +11:00
eulerzgy ea86e66fbe Del capatical local packagename for kubeletUtil 2015-10-15 09:58:10 +08:00
Tim Hockin 2f4c3035be Change to alpha/beta for bandwidth annotations 2015-10-14 15:33:19 -07:00
Wojciech Tyczynski 647aa1bc8c Unify per-resource List for unversioned client 2015-10-14 08:37:57 +02:00
Yu-Ju Hong 2c76c55bb9 kubelet: move PodManager and MirrorClient to a subpackage
This change moves pod_manager.go and mirror_client.go to a separate package.
Also made necessary, minor changes to facilitate testing.
2015-10-12 16:34:18 -07:00
Yu-Ju Hong 1c015560be kubelet logs update source in the sync loop
This improve the clarity/debuggability.
2015-10-12 10:18:52 -07:00
Mike Danese 392f33e417 Merge pull request #14054 from mikedanese/register-master
register master kubelet with the apiserver
2015-10-09 17:12:37 -07:00
Yu-Ju Hong a3e60cc32e Rename imported package local name kubeletTypes to kubetypes
According to the naming guidelines, package name should not include mixedCaps.
2015-10-09 10:24:31 -07:00
k8s-merge-robot 45cca144cb Merge pull request #15167 from yujuhong/sources_seen
Auto commit by PR queue bot
2015-10-09 08:35:40 -07:00
k8s-merge-robot b793c3edf1 Merge pull request #14700 from liggitt/kubelet_authz
Auto commit by PR queue bot
2015-10-09 03:01:43 -07:00
k8s-merge-robot 538cf72208 Merge pull request #15051 from feiskyer/kubelet/garbage-collection
Auto commit by PR queue bot
2015-10-09 02:11:47 -07:00
Jordan Liggitt 9d6b52881d Add authentication/authorization interfaces to kubelet, always include /metrics with /stats 2015-10-09 03:10:00 -04:00
Mike Danese fa60bbe8e6 add flag to kubelet to ignore the cidr passed down by the apiserver on the master 2015-10-08 21:21:19 -07:00
feisky fb04edea3a Replace rkt --grace-period and --expire-prepared with gcPolicy.MinAge 2015-10-09 06:57:10 +08:00
feisky d624c7de51 Pass the ContainerGCPolicy in Runtime.GarbageCollect 2015-10-09 06:57:10 +08:00
feisky 35cf8ab7d4 Use runtime instread of dockerclient in container gc 2015-10-09 06:54:38 +08:00
Yu-Ju Hong 098ab05997 kubelet: move common types to kubelet/types
This would faciliate tasks such as moving code in pkg/kubelet to sub packages.
2015-10-08 14:38:01 -07:00
Piotr Szczesniak 95b293c615 Merge pull request #13516 from samsabed/backoff-image-reason
backoff image pulling on failure
2015-10-08 13:05:32 +02:00
Tim St. Clair 551eff63b8 Use strong type for container ID
Change all references to the container ID in pkg/kubelet/... to the
strong type defined in pkg/kubelet/container: ContainerID

The motivation for this change is to make the format of the ID
unambiguous, specifically whether or not it includes the runtime
prefix (e.g. "docker://").
2015-10-07 10:58:05 -07:00
Derek Carr 87fe1f4af8 Merge pull request #13939 from pmorie/pod-sc-minimal
Add PodSecurityContext and backward compatibility tests
2015-10-07 09:56:58 -04:00
Sam Abed fdf712cd84 back-off image pull on failure
Signed-off-by: Sam Abed <samabed@gmail.com>
2015-10-07 21:12:42 +11:00
Yu-Ju Hong 5abdfcdfe6 kubelet: fix all sources ready condition
The current implementation considers a source seen when it receives a SET at
kubelet/config/config.go. However, the main kubelet sync loop may not have
received the pod update from the source via the channel. This change ensures
that kubelet would consider all sources are ready only after the sync loop has
seen all the sources.
2015-10-06 10:13:04 -07:00
Paul Morie 227dd82119 Add PodSecurityContext and backward compatibility tests 2015-10-05 21:05:27 -04:00
Yifan Gu b42d2311c3 kubelet/rkt: add container/image gc for rkt. 2015-10-05 14:22:19 -07:00
Alex Robinson d8120f5425 Merge pull request #14221 from timstclair/readiness-workers
Refactor readiness probing
2015-10-05 13:26:54 -07:00
Tim St. Clair 52ece0c34e Refactor readiness probing
Each container with a readiness has an individual go-routine which
handles periodic probing for that container. The results are cached, and
written to the status.Manager in the pod sync path.
2015-10-02 15:37:10 -07:00
k8s-merge-robot 13382b30c4 Merge pull request #14950 from yujuhong/sync_type
Auto commit by PR queue bot
2015-10-02 12:39:00 -07:00
k8s-merge-robot 1d73382b83 Merge pull request #14958 from madhusudancs/fix-kubelet-messages
Auto commit by PR queue bot
2015-10-02 11:55:42 -07:00
k8s-merge-robot 6ca515c312 Merge pull request #14820 from gmarek/playground
Auto commit by PR queue bot
2015-10-02 07:37:45 -07:00
Madhusudan.C.S 7a709b09c3 Messages should be appended to "messages" slice.
Network configuration error message while setting Kubelet status was
being written to "reasons" slice. Write this message to "messages" slice
instead.

Also remove "reasons" slice entirely since it is not used anywhere.
2015-10-01 17:55:45 -07:00
Yu-Ju Hong 889e798ddb kubelet: pipe SyncPodType to pod workers
Now that kubelet has switched to incremental updates, it has complete
information of the pod update type (create, update, sync). This change pipes
this information to pod workers so that they don't have to derive the type
again.
2015-10-01 16:29:46 -07:00
k8s-merge-robot b19837f4e2 Merge pull request #14767 from nikhiljindal/podConditionTransitionTime
Auto commit by PR queue bot
2015-10-01 13:01:45 -07:00
gmarek fc79ce347e Allow usining FakeOOMAdjuster in Kubelet 2015-10-01 10:55:08 +02:00
k8s-merge-robot b661cfd21e Merge pull request #14643 from liguangbo/change_Oom_to_OOM
Auto commit by PR queue bot
2015-09-29 18:54:22 -07:00
nikhiljindal b99d225d19 Setting PodReadyCondition.LastTransitionTime 2015-09-29 17:23:32 -07:00
liguangbo 12299087da Change Oom to OOM 2015-09-28 18:18:04 +08:00
feisky d25ae52d9b Refactor image manager for client/server implementation of the container runtime 2015-09-26 08:29:08 +08:00
Brian Grant efd19143f7 Revert "Refactor image manager for client/server implementation of the container runtime" 2015-09-25 15:24:18 -07:00
Brian Grant bbc90cd9ff Merge pull request #14035 from feiskyer/kubelet/client/server
Refactor image manager for client/server implementation of the container runtime
2015-09-25 13:18:53 -07:00
Vishnu kannan 77be271ed0 Linux container manager initialization will now fail if cpu and memory cgroups are not mounted. 2015-09-24 15:34:42 -07:00
gmarek 66ca7fcc92 Functionally revert #12894 2015-09-23 12:15:38 +02:00
Dawn Chen a010e91913 Merge pull request #14292 from yujuhong/revert_node
Revert node status manager
2015-09-21 23:13:18 -07:00
Clayton Coleman c2e90cd154 Support extended pod logging options
Increase the supported controls on pod logging. Add validaiton to pod
log options. Ensure the Kubelet is using a consistent, structured way to
process pod log arguments.

Add ?sinceSeconds=<durationInSeconds>, &sinceTime=<RFC3339>, ?timestamps=<bool>,
?tailLines=<number>, and ?limitBytes=<number>
2015-09-21 15:39:32 -04:00
Yu-Ju Hong 077158ef86 Rebase to use unversioned api 2015-09-21 11:14:14 -07:00
Yu-Ju Hong cadc279505 Revert "kubelet: move all node status related methods to a separate file"
This reverts commit e7d1e47f31.
2015-09-21 11:06:38 -07:00
Yu-Ju Hong d7b9f8905f Revert "Move node information retreival to nodeManager"
This reverts commit 2369cd6401.
2015-09-21 11:00:04 -07:00
nikhiljindal 6523ec142b Adding more fields to PodCondition 2015-09-18 05:26:13 -07:00
Daniel Martí 586cb9126a Move pkg/util.Time to pkg/api/unversioned.Time
Along with our time.Duration wrapper, as suggested by @lavalamp.
2015-09-17 17:51:27 -07:00
k8s-merge-robot c1eb1a1b2f Merge pull request #13791 from fgrzadkowski/fix_secrets_in_docker
Auto commit by PR queue bot
2015-09-17 15:13:29 -07:00
k8s-merge-robot 7cee028497 Merge pull request #13778 from smarterclayton/unready_endpoints
Auto commit by PR queue bot
2015-09-17 11:36:17 -07:00
feisky f1124eb0b1 Refactor image manager to operate on Runtime 2015-09-17 23:02:00 +08:00
Filip Grzadkowski 7fe34f2050 Introduce nsenter writer for volume plugins 2015-09-17 16:40:29 +02:00
Clayton Coleman bf2decce81 Add NotReadyAddresses to Endpoints
In many cases clients may wish to view not ready addresses for endpoints
in order to do set membership prior to a pod being ready. For instance,
a pod that uses the service endpoints to connect to other pods under
the same service, but does not want to signal ready before it has
contacted at least a minimal number of other pods.

This is backwards compatible with old servers and clients. There is
an additional cost in size of endpoints before services ramp up, which
will add minor CPU and memory use for services that have a significant
number of pods which have not become ready.
2015-09-17 09:41:56 -04:00
Yu-Ju Hong 2369cd6401 Move node information retreival to nodeManager
nodeManager should handle most node object interaction with apiserver. This
moves exsiting node-watching and GetNodes() methods to nodeManager.
2015-09-16 17:17:42 -07:00
Yu-Ju Hong e7d1e47f31 kubelet: move all node status related methods to a separate file
The methods for registering a node and syncing node status to the apiserver
have grown large enough that it makes sense for them to live in a separate
place. This change adds a nodeManager to handle such interaction with the
apiserver.
2015-09-16 17:17:42 -07:00
k8s-merge-robot 4c46bc3243 Merge pull request #13989 from yujuhong/sync_first
Auto commit by PR queue bot
2015-09-16 17:16:23 -07:00
k8s-merge-robot f6fb0a58b6 Merge pull request #13874 from timstclair/status-manager
Auto commit by PR queue bot
2015-09-16 00:46:01 -07:00
Yu-Ju Hong 8f8541b4cd kubelet: make sure sync loop is run before performing housekeeping
Otherwise, kubelet may kill the existing pods before having a chance to update
the desired pods.
2015-09-15 13:11:55 -07:00
gmarek 4b8ddf3d7e Add information about ports opened by Kubelet to API 2015-09-15 14:24:55 -04:00
Wojciech Tyczynski 5da52558e6 Fix watch with RV=0 in Kubelet. 2015-09-14 16:45:58 +02:00
k8s-merge-robot 6f01200188 Merge pull request #13727 from jiangyaoguo/fix-reason-reporting-in-kubelet
Auto commit by PR queue bot
2015-09-11 12:32:20 -07:00
Tim St. Clair 1f91fffb57 Move kubelet.statusManager to status.Manager
This refactor is in preparation for moving more state handling to the
status manager. It will become the canonical cache for the latest
information on running containers and probe status, as part of the
prober refactoring.
2015-09-11 12:22:01 -07:00
k8s-merge-robot bfc60709b1 Merge pull request #13468 from yifan-gu/rkt_doc
Auto commit by PR queue bot
2015-09-10 20:32:08 -07:00
jiangyaoguo 62c0c35307 Fix reporting reason in kubelet
1. Make reason field of StatusReport objects in kubelet in CamelCase format.
2. Add Message field for ContainerStateWaiting to describe detail about Reason.
3. Make reason field of Events in kubelet in CamelCase format.
4. Update swagger,deep-copy and so on.
2015-09-11 10:01:08 +08:00
Daniel Smith b225c1d47a Run gofmt (separate commit for easy rebases) 2015-09-10 17:17:59 -07:00
Daniel Smith 15b30b8b09 Move version agnostic parts of client
pkg/client/unversioned/cache -> pkg/client/cache
pkg/client/unversioned/record -> pkg/client/record
2015-09-10 17:17:59 -07:00
k8s-merge-robot ee7168d868 Merge pull request #13815 from robertabbott/abbott/kubelet_get_instead_of_list
Auto commit by PR queue bot
2015-09-10 15:59:01 -07:00
Yifan Gu e7b9c50ba1 kubelet/rkt: add support for different stage1 image.
Also enable grace stop for service files.
2015-09-10 14:58:03 -07:00
tummychow 78ce5da988 Move util.StringSet into its own package
A lot of packages use StringSet, but they don't use anything else from
the util package. Moving StringSet into another package will shrink
their dependency trees significantly.
2015-09-10 12:04:15 -07:00
Bobby Abbott 9a5f43e16d Kubelet#GetNode now uses GetNodeInfo instead of List
Both GetNode and the cache.ListWatch listfunc in the
kubelet package call List unnecessary.

GetNodeInfo is sufficient for GetNode and makes looping
through a list of nodes to check for a matching name
unnecessary.

resolves #13476
2015-09-10 10:50:41 -07:00
Chao Xu 002dcc6343 Merge pull request #13737 from wojtek-t/switch_to_node_in_registry
Rename "minion" to "node" in few places.
2015-09-09 17:49:22 -07:00
Wojciech Tyczynski cd1ac360de Rename "minion" to "node" in few places. 2015-09-09 16:18:23 +02:00
jiangyaoguo 1460a1fb9e Rate limit events in kubelet
1. Add EvnetRecordQps and EventBurst parameter in kubelet.
2. If  EvnetRecordQps and EventBurst was set, rate limit events in kubelet
with a independent ratelimiter as setted.
2015-09-09 16:57:21 +08:00
feihujiang d713826aa9 Add pod name to error message when getting logs of a container 2015-09-07 17:03:54 +08:00
James DeFelice de064f4254 fix kubelet syncLoop busy loop upon close of updates chan 2015-09-04 17:56:02 +00:00
Abhi Shah c9f2d3151b Merge pull request #13155 from derekwaynecarr/enforce_limits
Map cpu limits to cpu-quota in docker run
2015-09-03 16:28:04 -07:00
Abhi Shah d923e61dda Merge pull request #13397 from yujuhong/housekeeping
kubelet: define the housekeeping period
2015-09-03 15:46:29 -07:00
derekwaynecarr 5dc74e8dbf Add support for CFS quota in kubelet 2015-09-03 13:44:28 -04:00
Brendan Burns ac958e5320 Merge pull request #12513 from brendandburns/tc2
integrate bandwidth shaping and the kubelet.
2015-09-02 09:44:10 -07:00
Yu-Ju Hong 08ec796995 kubelet: remove orphaned pod directories properly
PR #13293 added a safety check to not remove a pod directory if the child
volumes directory is not empty. This logic is faulty because kubelet may have
directory structures podUID/volumes/volumeKind/volumeName. E.g.,
`056db95d-50ee-11e5-a2e4-42010af0ba1d/volumes/kubernetes.io~empty-dir/default-token-al3r2`

This change fixes that by properly listing all volumes under a pod.
2015-09-01 15:39:17 -07:00
k8s-merge-robot 5331a26c5b Merge pull request #13293 from yujuhong/pod_dir
Auto commit by PR queue bot
2015-09-01 04:59:55 -07:00
Yu-Ju Hong 13b268b378 kubelet: define the minimum housekeeping period
Before, kubelet performs global cleanup tasks every iteration. After the
PR #13003, kubelet performs the tasks on every sync internval (10 seconds).
This PR decouples the housekeeping period with the sync internval to ensure
that kubelet cleans up promptly, while not too often (no more than once every
minimum housekeeping period).
2015-08-31 17:28:20 -07:00
Brendan Burns 9f3ef68ebc integrate bandwidth shaping and the kubelet. 2015-08-30 22:24:09 -07:00
k8s-merge-robot 4a7115d272 Merge pull request #11955 from pedro-r-marques/flag_resolv_conf
Auto commit by PR queue bot
2015-08-30 21:03:54 -07:00
Yu-Ju Hong 8774f6efa2 kubelet: don't remove any pod directory with non-empty volumes
If the volumes directory is not empty, it implies that volumes may not have
been properly cleaned up yet. Do not attempt to remove the pod directory.
2015-08-27 22:18:51 -07:00
Yu-Ju Hong 697197d383 Kubelet: do not remove pod directory if any container is still running
If there are any running container in the pod, we cannot remove the volume.
Therefore, we should not attempt to remove the pod directory.
2015-08-27 18:19:13 -07:00
Pedro Roque Marques 99b1da848d Add kubelet '--resolv-conf' flag.
Allow the user to specify the resolver configuration file that is used
to determine the default DNS parameters. This defaults to the system's
/etc/resolv.conf.
2015-08-26 15:41:36 -07:00
Yu-Ju Hong c7afae75f5 Merge pull request #13103 from yujuhong/restart_count
Kubelet: fix container restart counts
2015-08-26 15:13:10 -07:00
Yu-Ju Hong b906e34576 kubelet: trigger pod workers independently
Currently, whenever there is any update, kubelet would force all pod workers to
sync again, causing resource contention and hence performance degradation.

This commit flips kubelet to use incremental updates (as opposed to snapshots).
This allows us to know what pods have changed and send updates to those pod
workers only. The `SyncPods` function has been replaced with individual
handlers, each handling an operation (ADD, REMOVE, UPDATE). Pod workers are
still triggered periodically, and kubelet performs periodic cleanup as well.

This commit also spawns a new goroutine solely responsible for killing pods.
This is necessary because pod killing could hold up the sync loop for
indefinitely long amount of time now user can define the graceful termination
period in the container spec.
2015-08-25 17:52:01 -07:00
Yu-Ju Hong 9f508e9422 Merge pull request #13076 from invenfantasy/master
Cleanup deprecated Forever function
2015-08-25 10:07:10 -07:00
Nikhil Jindal 8846c841a8 Merge pull request #12648 from samsabed/crashloop
back off restarts of crashlooping containers
2015-08-24 18:54:13 -07:00
Yu-Ju Hong 1e26cdfb67 kubelet: fix container restart counts
This commit ensures that the container restart count increments monotonically
when there is no kubelet restarts.
2015-08-24 12:50:05 -07:00
Nikhil Jindal a1c5f3f45d Merge pull request #12876 from yujuhong/pod_status
kubelet: switch to using pod UID as the key in status manager
2015-08-24 10:39:02 -07:00
Sam Abed 995cb15bb6 back off restarts of crashlooping containers
Signed-off-by: Sam Abed <samabed@gmail.com>
2015-08-24 21:43:17 +10:00
caogaojin 60cb4406e9 Cleanup deprecated Forever function
Since util.Forever function has been deprecated, we should cleanup these
pieces of code.
2015-08-24 10:31:59 +08:00
Clayton Coleman f5c4a3e7a6 Handle graceful termination in the Kubelet
Supports pods that are gracefully deleted on the server being
handled in the Kubelet

preStop is limited to the grace period of the pod in execution.
2015-08-21 10:18:21 -04:00
Saad Ali 9b01580946 Merge pull request #12872 from runningwild/master
Fix several problems with using rkt on gce
2015-08-20 15:46:16 -07:00
Jonathan Wills 80e799fc0c Allow configuring the rkt binary in a kubelet with a flag.
This is necessary because coreos comes with rkt installed, and if we want to use a different version
we need some way to avoid the default one.
2015-08-19 09:42:13 -04:00
Robert Bailey 08e6a43c1d Revert "Merge pull request #9165 from smarterclayton/graceful"
This reverts commit 4f856b595d, reversing
changes made to d78525a83b.

Conflicts:
	pkg/kubelet/status_manager.go
2015-08-18 17:34:49 -07:00
Yu-Ju Hong 0c84b837cf kubelet: switch to using pod UID as the key in status manager
We chose to use podFullName (name_namespace) as key in the status manager
because mirror pod and static pod share the same status. This is no longer
needed because we do not store statuses for static pods anymore (we only
store statuses for their mirror pods). Also, reviously, a few fixes were
merged to ensure statuses are cleaned up so that a new pod with the same
name would not resuse an old status.

This change cleans up the code by using UID as key so that the code would
become less brittle.
2015-08-18 14:39:02 -07:00
Robert Bailey 11d5712278 Merge pull request #12736 from deads2k/relax-pull-secret-constraints
tolerate missing image pull secrets:
2015-08-18 11:48:01 -07:00
Kris Rousey ae6c64d9bb Moving everyone to unversioned client 2015-08-18 10:23:03 -07:00
Clayton Coleman 780accb3ba Kubelet should garbage collect dead pods
The sync loop should check for terminated pods that are no longer
running and clear them. The status loop should never write status
if the pod UID changes. Mirror pods should be deleted immediately
rather than gracefully.
2015-08-18 09:08:44 -04:00
Clayton Coleman 89f1f3b1b8 Alter graceful deletion to not use TTL
Avoid TTL by deleting pods immediately when they aren't
scheduled, and letting the Kubelet delete them otherwise.

Ensure the Kubelet uses pod.Spec.TerminationGracePeriodSeconds
when no pod.DeletionGracePeriodSeconds is available.
2015-08-18 09:08:43 -04:00
Clayton Coleman b842a7dd15 Revert "Revert "Gracefully delete pods from the Kubelet""
This reverts commit 98115facfd.
2015-08-18 08:57:02 -04:00
Clayton Coleman 01f3785426 Add status.podIP as a valid downward API target
Getting the public IP a container is supposed to use is O(hard),
and usually involves ugly gyrations in python or with interfaces.
Using the downward API means that the IP Kube is announcing to
other endpoints is also visible inside the container for pods to
identify themselves.
2015-08-14 17:51:55 -04:00
deads2k c24d176f65 tolerate missing image pull secrets: 2015-08-14 12:51:28 -04:00
Yifan Gu d70a30c069 kubelet: refactor kubelet.Runtimehooks to container.ImagePuller. 2015-08-12 16:28:25 -07:00
jiangyaoguo 9ab4a46b9d Keep event reason in kubelet consistent with others 2015-08-12 19:00:09 +08:00
Piotr Szczesniak 1df0267f4a Merge pull request #12551 from eparis/underscore-to-dash
Update code and docs to use - in flag names instead of _
2015-08-12 07:16:31 +02:00
Yu-Ju Hong 56f4605f47 kubelet: refactor SyncPods for better readability
Eventually we would like to replace the all-encompassing SyncPods function with
more well-defined, smaller functions. This would not only help with the
readability and profiling of the code, it'd also set in motion for the plans to
trigger pod worker individually based on the content of the pod updates.

This commit serves as the first step of that, while avoiding breaking all unit
tests by preserving the SyncPods function for the time being.
2015-08-11 16:40:26 -07:00
Kris Rousey 565189f5b8 Correcting all go vet errors 2015-08-11 13:55:37 -07:00
Eric Paris 5aa495cdad Update code to use - in flag names instead of _ 2015-08-11 16:31:52 -04:00
Marek Grabowski 5f9cefc1d8 Merge pull request #12441 from vlajos/typofixes-vlajos-20150807
typofix - https://github.com/vlajos/misspell_fixer
2015-08-10 16:33:52 +02:00
Marek Grabowski 8580e17ff1 Merge pull request #12374 from wulonghui/wulonghui-patch-1
Add NodeInternalIP for Node's Addresses when cloud provider is nil
2015-08-10 15:01:00 +02:00
Marek Grabowski e034712456 Merge pull request #11788 from HaiyangDING/ImproveClarityResource
Improve clarity around PodFitsResource(issue#11453)
2015-08-10 09:56:32 +02:00
Veres Lajos 9f77e49109 typofix - https://github.com/vlajos/misspell_fixer 2015-08-08 22:31:48 +01:00
Ananya Kumar 6ef3de1d5f Add QoS support on node 2015-08-07 11:18:16 -07:00
dinghaiyang dab7280ae4 Improve clarity around PodFitsResource by showing pods limits in `kubectl describe node` 2015-08-07 16:58:11 +08:00
wulonghui b50e736184 Add NodeInternalIP for Node's Addresses when cloud provider is nil 2015-08-07 09:08:56 +08:00
Mike Danese fe6b15ba2f rewrite all links to issues to k8s links 2015-08-05 21:11:11 -07:00