Commit Graph

546 Commits (1faa6f56b9c026b58d17e9163bf0b26a8b4086b4)

Author SHA1 Message Date
m1093782566 1faa6f56b9 fix kube-proxy panic 2017-09-01 16:18:03 +08:00
Kubernetes Submit Queue b65f3cc8dd Merge pull request #49850 from m1093782566/service-session-timeout
Automatic merge from submit-queue (batch tested with PRs 49850, 47782, 50595, 50730, 51341)

Paramaterize `stickyMaxAgeMinutes` for service in API

**What this PR does / why we need it**:

Currently I find `stickyMaxAgeMinutes` for a session affinity type service is hard code to 180min. There is a TODO comment, see

https://github.com/kubernetes/kubernetes/blob/master/pkg/proxy/iptables/proxier.go#L205

I think the seesion sticky max time varies from service to service and users may not aware of it since it's hard coded in all proxier.go - iptables, userspace and winuserspace.

Once we parameterize it in API, users can set/get the values for their different services.

Perhaps, we can introduce a new field `api.ClientIPAffinityConfig` in `api.ServiceSpec`.

There is an initial discussion about it in sig-network group. See,

https://groups.google.com/forum/#!topic/kubernetes-sig-network/i-LkeHrjs80

**Which issue this PR fixes**: 

fixes #49831

**Special notes for your reviewer**:

**Release note**:

```release-note
Paramaterize session affinity timeout seconds in service API for Client IP based session affinity.
```
2017-08-25 20:43:30 -07:00
m1093782566 c355a2ac96 Paramaterize stickyMaxAgeMinutes for service in API 2017-08-25 17:44:47 +08:00
m1093782566 a7fd545d49 clean up LocalPort in proxier.go 2017-08-24 11:16:38 +08:00
m1093782566 992f618343 move IsLocalIP() and ShouldSkipService() to pkg/proxy/util 2017-08-17 14:32:05 +08:00
Kubernetes Submit Queue 41784b5c66 Merge pull request #49478 from m1093782566/proxy-util-conntrack
Automatic merge from submit-queue (batch tested with PRs 50094, 48966, 49478, 50593, 49140)

[kube-proxy] Move UDP conntrack operations together to pkg/proxy/util/conntrack.go

**What this PR does / why we need it**:

Fix TODO in pkg/proxy/iptables.go, see

https://github.com/kubernetes/kubernetes/blob/master/pkg/proxy/iptables/proxier.go#L1632

Move UDP conntrack operations together to from `pkg/proxy/iptables/proxier.go` to `pkg/proxy/util/conntrack.go` so that make them more consistent and add some UTs.

**Which issue this PR fixes**

Fixes #49477

**Special notes for your reviewer**:

```release-note
NONE
```
2017-08-14 12:14:59 -07:00
Kubernetes Submit Queue 2d4c609a1d Merge pull request #49847 from m1093782566/fix-win-proxy
Automatic merge from submit-queue

Fix winspace proxier wrong comment message

**What this PR does / why we need it**:

Since winspace proxier has nothing to do with iptables, this PR remove the wrong comment message on iptables.

**Which issue this PR fixes**: 

Fixes #50524
2017-08-12 01:51:07 -07:00
m1093782566 7b8372db99 move UDP conntrack operations together to pkg/proxy/util/conntrack.go 2017-08-12 11:10:04 +08:00
Jeff Grafton a7f49c906d Use buildozer to delete licenses() rules except under third_party/ 2017-08-11 09:32:39 -07:00
Jeff Grafton 33276f06be Use buildozer to remove deprecated automanaged tags 2017-08-11 09:31:50 -07:00
Jeff Grafton cf55f9ed45 Autogenerate BUILD files 2017-08-11 09:30:23 -07:00
xiangpengzhao ea1a577358 Remove some helpers associated with ESIPP. 2017-08-09 14:25:08 +08:00
xiangpengzhao ebe21ee4c1 Remove deprecated ESIPP beta annotations 2017-08-05 15:00:58 +08:00
Kubernetes Submit Queue 5d24a2c199 Merge pull request #49300 from tklauser/syscall-to-x-sys-unix
Automatic merge from submit-queue

Switch from package syscall to golang.org/x/sys/unix

**What this PR does / why we need it**:

The syscall package is locked down and the comment in https://github.com/golang/go/blob/master/src/syscall/syscall.go#L21-L24 advises to switch code to use the corresponding package from golang.org/x/sys. This PR does so and replaces usage of package syscall with package golang.org/x/sys/unix where applicable. This will also allow to get updates and fixes
without having to use a new go version.

In order to get the latest functionality, golang.org/x/sys/ is re-vendored. This also allows to use Eventfd() from this package instead of calling the eventfd() C function.

**Special notes for your reviewer**:

This follows previous works in other Go projects, see e.g. moby/moby#33399, cilium/cilium#588

**Release note**:

```release-note
NONE
```
2017-08-03 04:02:12 -07:00
Kubernetes Submit Queue 7b55a2d385 Merge pull request #48624 from derekwaynecarr/log-abridged
Automatic merge from submit-queue

Log abridged set of rules at v2 in kube-proxy on error

**What this PR does / why we need it**:
this is a follow-on to https://github.com/kubernetes/kubernetes/pull/48085

**Special notes for your reviewer**:
we hit this in operations where we typically run in v2, and would like to log abridged set of output rather than full output.

**Release note**:
```release-note
NONE
```
2017-08-02 15:26:37 -07:00
m1093782566 d56344fdf7 fix winspace wrong comment message 2017-08-01 14:06:23 +08:00
Zihong Zheng db379de778 Emit event and retry when fail to start healthz server on kube-proxy. 2017-07-27 20:54:51 -07:00
Derek Carr 3bddef790b Log abridged set of rules at v2 in kube-proxy on error 2017-07-25 10:32:16 -04:00
supereagle adc0eef43e remove duplicated import and wrong alias name of api package 2017-07-25 10:04:25 +08:00
Tobias Klauser 4a69005fa1 switch from package syscall to x/sys/unix
The syscall package is locked down and the comment in [1] advises to
switch code to use the corresponding package from golang.org/x/sys. Do
so and replace usage of package syscall with package
golang.org/x/sys/unix where applicable.

  [1] https://github.com/golang/go/blob/master/src/syscall/syscall.go#L21-L24

This will also allow to get updates and fixes for syscall wrappers
without having to use a new go version.

Errno, Signal and SysProcAttr aren't changed as they haven't been
implemented in /x/sys/. Stat_t from syscall is used if standard library
packages (e.g. os) require it. syscall.SIGTERM is used for
cross-platform files.
2017-07-21 12:14:42 +02:00
ymqytw 9b393a83d4 update godep 2017-07-20 11:03:49 -07:00
ymqytw 3dfc8bf7f3 update import 2017-07-20 11:03:49 -07:00
Mike Danese c201553f27 remove some people from OWNERS so they don't get reviews anymore
These are googlers who don't work on the project anymore but are still
getting reviews assigned to them:
- bprashanth
- rjnagal
- vmarmol
2017-07-13 10:02:21 -07:00
Kubernetes Submit Queue 8c52c8f15c Merge pull request #48635 from dcbw/userspace-proxy-silence-loadbalancerrr-message
Automatic merge from submit-queue (batch tested with PRs 48425, 41680, 48457, 48619, 48635)

proxy/userspace: suppress "LoadBalancerRR: Removing endpoints" message

Don't print it when there aren't any endpoints getting removed.

Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1468420
Fixes: https://github.com/kubernetes/kubernetes/issues/48816

```release-note
NONE
```

@eparis @thockin @kubernetes/rh-networking
2017-07-12 10:57:21 -07:00
Kubernetes Submit Queue 923ef03d05 Merge pull request #48613 from dcbw/userspace-proxy-host-ip
Automatic merge from submit-queue (batch tested with PRs 47232, 48625, 48613, 48567, 39173)

proxy/userspace: honor listen IP address as host IP if given

Allows the proxier to be used on an interface that's not the default route,
otherwise hostIP gets set to the default route interface even if that's
not what the user intended.

If listen IP isn't given, falls back to previous behavior.

```release-note
To allow the userspace proxy to work correctly on multi-interface hosts when using the non-default-route interface, you may now set the `bindAddress` configuration option to an IP address assigned to a network interface.  The proxy will use that IP address for any required NAT operations instead of the IP address of the interface which has the default route.
```

@kubernetes/sig-network-misc @thockin @wojtek-t
2017-07-12 00:10:16 -07:00
Cao Shufeng 0c577c47d5 Use glog.*f when a format string is passed
ref:
https://godoc.org/github.com/golang/glog

I use the following commands to search all the invalid usage:
$ grep "glog.Warning(" * -r | grep %
$ grep "glog.Info(" * -r | grep %
$ grep "glog.Error(" * -r | grep %
$ grep ").Info(" * -r | grep % | grep "glog.V("
2017-07-10 19:04:03 +08:00
Dan Williams 0233f2b924 proxy/userspace: suppress "LoadBalancerRR: Removing endpoints" message
Don't print it when there aren't any endpoints getting removed.

Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1468420
2017-07-07 15:45:30 -05:00
Dan Williams 8a58f62fe0 proxy/userspace: honor listen IP address as host IP if given
Allows the proxier to be used on an interface that's not the default route,
otherwise hostIP gets set to the default route interface even if that's
not what the user intended.

If listen IP isn't given, falls back to previous behavior.
2017-07-07 11:17:17 -05:00
Minhan Xia 68a2749b28 fix unit tests 2017-07-06 16:01:03 -07:00
Minhan Xia 46d3e83caf refactor updateEndpointMap and updateServiceMap results 2017-07-06 16:00:57 -07:00
Minhan Xia 25ac521f88 flush conntrack entry for udp service when # of backend changes from 0 to non-0 2017-07-06 15:12:44 -07:00
xiangpengzhao 9e31eb280a Populate endpoints and allow ports with headless service 2017-06-28 11:15:51 +08:00
Shyam Jeedigunta d354e0bf4f Move iptables logging in kubeproxy from Errorf to V(2).Infof 2017-06-26 19:59:23 +02:00
Chao Xu 60604f8818 run hack/update-all 2017-06-22 11:31:03 -07:00
Chao Xu f2d3220a11 run root-rewrite-import-client-go-api-types 2017-06-22 11:30:59 -07:00
Kubernetes Submit Queue 9054e70ca3 Merge pull request #38751 from chenchun/warn
Automatic merge from submit-queue

Raise a warning instead of info if br-netfilter is missing or unset

Took quite a while to figure out why service VIP is unreachable on my cluster. It turns out br-nf-call-iptables is unset. I wish this message could be a warning to attract considerable attention.
2017-06-16 17:25:32 -07:00
Tim Hockin ce8309780f Proxy: comments around ClusterCIDR use 2017-06-01 12:36:31 -07:00
Wojciech Tyczynski 03c255d7c5 Store chain names to avoid recomputing them multiple times 2017-05-30 10:50:10 +02:00
Wojciech Tyczynski c4d51f12a2 Store port endpoint chain names to avoid recomputing it multiple times 2017-05-30 10:49:36 +02:00
Wojciech Tyczynski 88e3e8f470 Reuse args slice for generating iptable rules. 2017-05-30 10:49:35 +02:00
Wojciech Tyczynski 070f393bc8 Precompute probabilities in iptables kube-proxy. 2017-05-30 10:49:34 +02:00
Wojciech Tyczynski 1242e8ca20 Minor improvement for memory allocations 2017-05-30 08:11:12 +02:00
Kubernetes Submit Queue 25dc892a97 Merge pull request #46450 from MrHohn/fix-proxy-healthcheck-nilpointer
Automatic merge from submit-queue (batch tested with PRs 46450, 46272, 46453, 46019, 46367)

Fix potential nil pointer dereference for kube-proxy healthcheck

Found error log from https://storage.googleapis.com/kubernetes-jenkins/logs/ci-kubernetes-e2e-gce-slow/5222:
```
I0524 20:22:35.554156       7 healthcheck.go:226] Not saving endpoints for unknown healthcheck "kube-system/kubernetes-dashboard"
I0524 20:22:35.554172       7 proxier.go:923] syncProxyRules took 40.047209ms
I0524 20:22:35.554218       7 healthcheck.go:175] Healthcheck "e2e-tests-esipp-f7djn/external-local" closed: accept tcp [::]:32027: use of closed network connection
I0524 20:22:37.416133       7 logs.go:41] http: panic serving 169.254.169.254:49216: runtime error: invalid memory address or nil pointer dereference
goroutine 1623 [running]:
net/http.(*conn).serve.func1(0xc420b27220)
	/usr/local/go_k8s_patched/src/net/http/server.go:1721 +0xd0
panic(0x1c07e40, 0x2b11ee0)
	/usr/local/go_k8s_patched/src/runtime/panic.go:489 +0x2cf
k8s.io/kubernetes/pkg/proxy/healthcheck.hcHandler.ServeHTTP(0xc4201aea60, 0x15, 0xc4203233a0, 0xe, 0xc4203aa280, 0x2ab51a0, 0xc420ae4d20, 0xc4202e7a00)
	/go/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/pkg/proxy/healthcheck/healthcheck.go:200 +0x86
k8s.io/kubernetes/pkg/proxy/healthcheck.(*hcHandler).ServeHTTP(0xc420798390, 0x2ab51a0, 0xc420ae4d20, 0xc4202e7a00)
	<autogenerated>:8 +0x87
net/http.serverHandler.ServeHTTP(0xc4208d0210, 0x2ab51a0, 0xc420ae4d20, 0xc4202e7a00)
	/usr/local/go_k8s_patched/src/net/http/server.go:2568 +0x92
net/http.(*conn).serve(0xc420b27220, 0x2ab66e0, 0xc420945380)
	/usr/local/go_k8s_patched/src/net/http/server.go:1825 +0x612
created by net/http.(*Server).Serve
	/usr/local/go_k8s_patched/src/net/http/server.go:2668 +0x2ce
```

So seems like it's possible healthcheck server is still serving traffic after service entry is removed.

/assign @freehan 

**Release note**:

```release-note
NONE
```
2017-05-26 18:48:56 -07:00
Zihong Zheng 41c4e965c3 Fix potential nil pointer dereference for kube-proxy healthcheck 2017-05-25 10:26:51 -07:00
Tim Hockin 2856fde23b Use BoundedFrequencyRunner in kube-proxy 2017-05-24 20:33:15 -07:00
Tim Hockin 578d9fcf63 Logging/naming cleanup for service port names 2017-05-24 20:33:15 -07:00
Kubernetes Submit Queue 6f5193593d Merge pull request #46201 from wojtek-t/address_kubeproxy_todos
Automatic merge from submit-queue

Address remaining TODOs in kube-proxy.

Followup PR from the previous two.
2017-05-22 20:54:14 -07:00
Wojciech Tyczynski 9e6de42745 Remove Save() from iptables interface 2017-05-22 13:23:49 +02:00
Wojciech Tyczynski ee505677f7 Address remaining TODOs in kube-proxy. 2017-05-22 09:18:55 +02:00
Kubernetes Submit Queue 336fb2f508 Merge pull request #45933 from smarterclayton/secret_reuse
Automatic merge from submit-queue

Move the remaining controllers to shared informers

Completes work done in 1.6 to move the last two hold outs to shared informers - tokens controller and scheduler. Adds a few more tools to allow informer reuse (like filtering the informer, or maintaining a mutation cache).

The mutation cache is identical to #45838 and will be removed when that merges

@ncdc @deads2k extracted from openshift/origin#14086
2017-05-20 23:08:09 -07:00