Commit Graph

203 Commits

Author SHA1 Message Date
Chao Xu
3688a78769 let load and density e2e tests use GC if it's on 2016-08-11 11:48:20 -07:00
Kubernetes Submit Queue
e7d01097dc Merge pull request #29971 from caesarxuchao/fix-kubectl-rolling-update-with-gc
Automatic merge from submit-queue

[GarbageCollector] Fix kubectl rolling-update to work with GC

This changes the order of the [Rename()](https://github.com/kubernetes/kubernetes/blob/master/pkg/kubectl/rolling_updater.go#L532) function. After the change, Rename() first deletes the old RC and orphans its pods, then creates the new RC, which will then have a chance to adopt the orphaned pods.

This also fixes the "should support rolling-update to same image" [test](https://github.com/kubernetes/kubernetes/blob/master/test/e2e/kubectl.go#L915) when the garbage collector is on.

Here is the detailed explanation on why the test would have failed:
`kubectl rolling-update` will [rename](https://github.com/kubernetes/kubernetes/blob/master/pkg/kubectl/rolling_updater.go#L532-L546) the RC. It first creates the an identical RC (including spec.selectors) with the new name, then it deletes the existing RC. When GC is turned on, the newly created RC cannot adopt the existing pod, because it has a controllerRef pointing to the exising RC, so the new RC will create new pods and expect to see the creation. However, the new RC and the old RC have the same selector, so sometimes the old RC, instead of the new RC, has its [expectation lowered](https://github.com/kubernetes/kubernetes/blob/master/pkg/controller/replication/replication_controller.go#L346-L362), the new RC's expectation will stuck forever. The e2e test then times out when executing `kubectl delete newRC`, because there is the new RC will not scale down as its expectation is not fulfilled.

A side-note, we should fix [rm.getPodController()](https://github.com/kubernetes/kubernetes/blob/master/pkg/controller/replication/replication_controller.go#L346) to respect pod's controllerref, that will prevent similar bugs.

Also note that an old version `kubectl rolling-update` will not work with the GC. We cannot fix that.
2016-08-04 20:48:10 -07:00
Chao Xu
96c84303bd fix kubectl rolling-update when GC is enabled 2016-08-03 15:57:03 -07:00
Wojciech Tyczynski
b456c6dbfd Increase test timeouts for creating LB in large clusters 2016-08-02 12:32:04 +02:00
Harry Zhang
c495397cae Refactor uuid into its own pkg 2016-07-30 00:07:02 -04:00
PingWang
1606ce57ea combine the const definition
Signed-off-by: PingWang <wang.ping5@zte.com.cn>
2016-07-26 20:28:12 +08:00
David McMahon
ef0c9f0c5b Remove "All rights reserved" from all the headers. 2016-06-29 17:47:36 -07:00
Wojciech Tyczynski
9961b8021b Fix failures in up and down services in large clusters 2016-06-15 16:09:46 +02:00
k8s-merge-robot
46323336cb Merge pull request #26559 from wojtek-t/max_idle_conns_per_host
Automatic merge from submit-queue

Max idle conns per host

Ref #26519

Hopefully, this will at least somehow help with #26519 

@gmarek @krousey @lavalamp
2016-06-01 11:41:42 -07:00
Wojciech Tyczynski
b67a6e6d41 Avoid creating multiple unnecessary clients in tests. 2016-05-31 14:57:09 +02:00
Jeff Lowdermilk
6ee2b7bc4e Fix some gce-only tests to run on gke as well
DaemonRestart kubelet test, and Services apiserver restart test.
2016-05-30 07:33:24 -07:00
Lucas Käldström
79ca1911e1 Removed all pause image constant strings, now the pause image is chosen by arch. Part of the effort of making e2e arch-agnostic 2016-05-26 19:16:43 +03:00
Daniel Smith
0bda969cd5 Merge pull request #25261 from jayunit100/readySched
Support GetReadySchedulable pods for all node selecting tests.
2016-05-18 17:24:09 -07:00
k8s-merge-robot
15919f336f Merge pull request #25284 from bprashanth/service_unready
Automatic merge from submit-queue

Add a service annotation that allows endpoints for unready pods

See https://github.com/kubernetes/kubernetes/issues/25283 for context 
@smarterclayton @thockin yes?

<!-- Reviewable:start -->
---
This change is [<img src="http://reviewable.k8s.io/review_button.svg" height="35" align="absmiddle" alt="Reviewable"/>](http://reviewable.k8s.io/reviews/kubernetes/kubernetes/25284)
<!-- Reviewable:end -->
2016-05-12 12:05:27 -07:00
k8s-merge-robot
5673f13edf Merge pull request #24014 from ihmccreery/chaosmonkey-no-ginkgo
Automatic merge from submit-queue

Upgrades tests use chaosmonkey package and ServiceTestJig

Introduce the `chaosmonkey` e2e package for doing disruptive testing (e.g. upgrade testing) more easily.

- [x] `chaosmonkey` package
- [x] migrate upgrade tests to `chaosmonkey` (using WIP `serviceJig`)
- [x] migrate upgrade tests to use `ServiceTestJig` and `chaosmonkey`

Deferred:

- [ ] make `ServiceTestJig` implement `chaosmonkey.Interface`
- [ ] migrate disruptive services tests to use `ServiceTestJig` and `chaosmonkey`

This provides the extensible framework for #15131.  We should now easily be able to add tests (e.g. #6084, #23189).

This is a rewrite of #22446.

cc @mikedanese @quinton-hoole @roberthbailey 

Assigning to @thockin, who wrote `ServiceTestJig`.

<!-- Reviewable:start -->
---
This change is [<img src="http://reviewable.k8s.io/review_button.svg" height="35" align="absmiddle" alt="Reviewable"/>](http://reviewable.k8s.io/reviews/kubernetes/kubernetes/24014)
<!-- Reviewable:end -->
2016-05-11 11:54:11 -07:00
Prashanth Balasubramanian
74ccd24574 Endpoints controller respects unready service annotation 2016-05-10 14:20:10 -07:00
Jay Vyas
326b213231 Consolidate node selecting tests to only use Schedulable + Running
nodes.
2016-05-06 10:32:12 -04:00
Muhammed Uluyol
f3690e2d5e
build/pause: write in C
Builds statically against glibc. References to the old pause
image have been updated.
2016-05-04 21:45:52 -04:00
Isaac Hollander McCreery
4e5e1b8ac5 Create chaosmonkey package, use ServiceTestJig for upgrade tests 2016-05-03 09:22:58 -07:00
Clayton Coleman
fdb110c859
Fix the rest of the code 2016-04-29 17:12:10 -04:00
Yu-Ju Hong
bdadaefe3b e2e: add a dummy environment variable in the service tests
This works around the docker bug:
https://github.com/docker/docker/issues/14203
2016-04-26 16:25:26 -07:00
Tim St. Clair
b0d3f32e88 Update test/e2e for test/e2e/framework refactoring 2016-04-13 10:50:17 -07:00
Jeff Lowdermilk
cd85b06465 Merge pull request #24001 from freehan/kubenet
include error in the fail msg
2016-04-08 20:55:35 -07:00
Minhan Xia
0174455111 include error in the fail msg 2016-04-07 15:42:21 -07:00
Joe Finney
39de4be40d
Mark services test slow. 2016-04-06 17:26:16 -07:00
Phillip Wittrock
f22ddf1ea8 Update serve_hostname version from 1.1 to v1.4 2016-03-21 12:53:22 -07:00
deads2k
ab03317d96 support CIDRs in NO_PROXY 2016-03-16 16:22:54 -04:00
Jay Vyas
9a71dfe753 KubeDescribe implementation with verify into after-build/ scripts. 2016-03-15 22:29:21 -04:00
Jeff Lowdermilk
f8d7793609 Merge pull request #22203 from bprashanth/network_flake
Ignore network flake in services e2e.
2016-03-03 15:20:12 -08:00
Prashanth Balasubramanian
41cbf952ca Ignore network flake in services e2e. 2016-02-29 17:40:24 -08:00
gmarek
110340c467 Add an option to pass client's QPS/burst to e2e framework 2016-02-29 09:32:29 +01:00
k8s-merge-robot
420d98fd8b Merge pull request #22027 from bprashanth/services_compare
Auto commit by PR queue bot
2016-02-26 17:25:22 -08:00
k8s-merge-robot
192119c2ac Merge pull request #21610 from fejta/e2etag
Auto commit by PR queue bot
2016-02-26 12:52:43 -08:00
Prashanth Balasubramanian
9be53e6319 Strip whitespace when comparing endpoints in services test. 2016-02-26 12:02:14 -08:00
k8s-merge-robot
f08a8f23c1 Merge pull request #20959 from justinsb/fix_20911
Auto commit by PR queue bot
2016-02-20 00:56:53 -08:00
Erick Fejta
3d8913a43a Use e2e tagged version of test images that aren't versioned 2016-02-19 20:22:49 -08:00
Tim Hockin
ff3f9f6bcc Move "extra" namespaces into framework
Now an aborted test will delete namespaces.  Hopefully this means less GCE
leaks.
2016-02-13 20:37:04 -08:00
Tim Hockin
677d4d6dbc e2e: Call AfterEach handlers in case of an abort
I was tired of aborted tests leaving debris in my cluster.
2016-02-13 20:29:46 -08:00
Justin Santa Barbara
46b89464fd e2e: Allow longer for AWS LoadBalancers to start serving traffic
When we create a LoadBalancer on AWS, there is a longer delay after
creating the LB before it starts to serve traffic than there is on GCE.
A delay of a few minutes is normal.

Use a longer timeout when waiting for the LB on AWS therefore.
2016-02-11 06:43:31 -05:00
Seth Jennings
982af896e1 use runId UUID for staticIPName to avoid collision 2016-02-10 21:42:17 -06:00
Justin Santa Barbara
a0093eb503 e2e: Don't try to create a UDP LoadBalancer on AWS
AWS doesn't support type=LoadBalancer with UDP services.  For now, we
simply skip over the test with type=LoadBalancer on AWS for the UDP
service.

Fix #20911
2016-02-10 12:31:18 -05:00
Yu-Ju Hong
4544575e11 e2e: use the tagged busybox image to avoid unnecessary pulling 2016-02-08 15:28:55 -08:00
k8s-merge-robot
14e2c2b740 Merge pull request #20773 from justinsb/fix/e2e_pods_to_be_ready_format
Auto commit by PR queue bot
2016-02-08 00:03:46 -08:00
Tim Hockin
fecb71420c Demote static IPs ASAP for easier cleanup
This exposed bugs in the IP promotion/demotion logic.
2016-02-06 21:15:06 -08:00
Justin Santa Barbara
051ff6aa35 e2e: fix format-string for error message
Argument was omitted
2016-02-06 09:15:55 -05:00
Tim Hockin
a98f556dbb cleaner service e2e strings 2016-02-04 09:28:48 -08:00
Saad Ali
b6c26ef69b Merge pull request #20569 from thockin/service-e2e-hang-20561
Fix wrong timeout param to wget
2016-02-03 16:00:13 -08:00
Saad Ali
ae35a302b5 Merge pull request #20471 from bprashanth/kp-logs
kube-proxy tries to apply latest snapshot
2016-02-03 15:49:02 -08:00
Tim Hockin
88dbcffdf0 Fix wrong timeout param to wget 2016-02-03 14:53:32 -08:00
Prashanth Balasubramanian
47f7f4417d Poll w/ timeout for nodeport to disappear. 2016-02-03 12:58:42 -08:00