github
/
k3s
mirror of https://github.com/k3s-io/k3s
1
0
Fork 0
Code Issues Projects Releases Wiki Activity
53,960 Commits
32 Branches
881 Tags
633 MiB
Commit Graph

147 Commits (ff471913f9073e8c69dd5c5e87fee95857a20aba)

Author SHA1 Message Date
David McMahon ef0c9f0c5b Remove "All rights reserved" from all the headers. 2016-06-29 17:47:36 -07:00
Sylwester Brzeczkowski af0f053cdf #2049 DeepCopy Service instead of shallow copy 2016-05-30 14:54:45 +02:00
Clayton Coleman c6961d6fd6
Remove runtime.Typer, reduce to ObjectKinds(1) (3) 
Remove the unnecessary variants, which avoids allocations in several
core paths.
 2016-05-21 23:55:42 -04:00
Jordan Liggitt 6c323a4f72 Remove name generation from thirdpartyresource 2016-05-09 09:27:54 -04:00
Clayton Coleman e0ebcf4216
Split the storage and negotiation parts of Codecs 
The codec factory should support two distinct interfaces - negotiating
for a serializer with a client, vs reading or writing data to a storage
form (etcd, disk, etc). Make the EncodeForVersion and DecodeToVersion
methods only take Encoder and Decoder, and slight refactoring elsewhere.

In the storage factory, use a content type to control what serializer to
pick, and use the universal deserializer. This ensures that storage can
read JSON (which might be from older objects) while only writing
protobuf. Add exceptions for those resources that may not be able to
write to protobuf (specifically third party resources, but potentially
others in the future).
 2016-05-05 12:08:23 -04:00
Timothy St. Clair 24b4286960 In preparation for new storage backends renaming generic registry store 2016-04-26 08:32:13 -05:00
k8s-merge-robot a275a045d1 Merge pull request #23914 from sky-uk/make-etcd-cache-size-configurable 
Automatic merge from submit-queue

Make etcd cache size configurable

Instead of the prior 50K limit, allow users to specify a more sensible size for their cluster.

I'm not sure what a sensible default is here. I'm still experimenting on my own clusters. 50 gives me a 270MB max footprint. 50K caused my apiserver to run out of memory as it exceeded >2GB. I believe that number is far too large for most people's use cases.

There are some other fundamental issues that I'm not addressing here:
- Old etcd items are cached and potentially never removed (it stores using modifiedIndex, and doesn't remove the old object when it gets updated)
- Cache isn't LRU, so there's no guarantee the cache remains hot. This makes its performance difficult to predict. More of an issue with a smaller cache size.
- 1.2 etcd entries seem to have a larger memory footprint (I never had an issue in 1.1, even though this cache existed there). I suspect that's due to image lists on the node status.

This is provided as a fix for #23323
 2016-04-17 00:06:31 -07:00
James Ravn 5bb0595260 Make deserialization cache size configurable 
Instead of the default 50K entries, allow users to specify more sensible
sizes for their cluster.
 2016-04-12 13:42:27 +01:00
Wojciech Tyczynski 53f433f019 Remove Set() from storage.Interface. 2016-04-04 17:54:18 +02:00
Wojciech Tyczynski 2699be2e7e Switch api.ObjetaMetaFor to meta.Accessor 2016-03-31 17:52:31 +02:00
AdoHe 5fdfc4bde3 fix can not export service bug 2016-03-05 11:23:50 -05:00
mqliang b0e06c14e5 add a knob to enable quorum read 2016-01-30 20:32:12 +08:00
Wojciech Tyczynski 58336829be Switch to versioned ListOptions in server. 2015-12-21 14:23:37 +01:00
deads2k 9fda7f1812 update StatusDetails to handle Groups 2015-12-17 09:14:12 -05:00
deads2k f764e0099c Update ObjectTyper to GroupVersion 2015-12-07 08:35:05 -05:00
Wojciech Tyczynski bf13bef955 Use unversioned.ListOptions in API server. 2015-11-24 14:42:21 +01:00
Timothy St. Clair 02851dd1b7 Removal of fakeClient and shift to storage.Interface for 
all registry tests.
 2015-11-19 10:34:30 -06:00
Wojciech Tyczynski 3df5d1dbc3 Move storage-related dirs under pkg/storage. 2015-11-12 19:49:32 +01:00
k8s-merge-robot cc53619372 Merge pull request #15791 from abutcher/set-service 
Auto commit by PR queue bot
 2015-10-30 06:55:16 -07:00
Andrew Butcher 0c4aafaf1e Update master service ports and type via controller. 2015-10-29 10:56:46 -04:00
Wojciech Tyczynski 9c53f301f1 Merge selectors with ListOptions. 2015-10-29 15:42:57 +01:00
Wojciech Tyczynski aa30e38183 Pass resource version to storage List operation. 2015-10-27 10:03:58 +01:00
Chao Xu f3caff2a79 testapi.Experimental -> testapi.Extensions 2015-10-09 15:15:35 -07:00
Wojciech Tyczynski cd1ac360de Rename "minion" to "node" in few places. 2015-09-09 16:18:23 +02:00
Chao Xu 9fc79e9d99 refactor testapi and test scripts to prepare for multiple API groups. 2015-09-04 18:01:32 -07:00
Wojciech Tyczynski 97b4a64e2c Refactor registry etcd delete tests 2015-09-03 09:05:17 +02:00
Wojciech Tyczynski a4baf4c211 Add registrytest support for etcd tests. 2015-09-02 07:54:34 +02:00
Wojciech Tyczynski 6dfe5b4b5a Refactoring of watch etcd tests. 2015-09-01 09:24:16 +02:00
nikhiljindal 25f2b85041 Deleting unused code 2015-08-27 19:52:47 -07:00
Wojciech Tyczynski 61de4fc34d Refactoring of create etcd tests. 2015-08-24 15:11:00 +02:00
Wojciech Tyczynski 5931cc74cc Unification refactoring to pkg/registry 2015-08-20 11:17:23 +02:00
Wojciech Tyczynski dcc302ab38 Refactoring of list etcd tests. 2015-08-19 13:56:23 +02:00
Wojciech Tyczynski 2ced08358f Remove unused registry infrastructure 2015-08-12 11:23:10 +02:00
Wojciech Tyczynski 79125f460c Services using standard REST storage 2015-08-10 13:27:44 +02:00
Wojciech Tyczynski d11ab96446 Extract non-storage operations from service etcd 2015-08-10 13:27:44 +02:00
Mike Danese 8e33cbfa28 rewrite go imports 2015-08-05 17:30:03 -07:00
hurf 6203ce9f85 Move pkg/scheduler to plugin/pkg/scheduler 
As the TODO in plugin/pkg/scheduler/scheduler.go described:

move everything from pkg/scheduler into this package. Remove
references from registry.
 2015-05-13 14:03:25 +08:00
Eric Paris 6b3a6e6b98 Make copyright ownership statement generic 
Instead of saying "Google Inc." (which is not always correct) say "The
Kubernetes Authors", which is generic.
 2015-05-01 17:49:56 -04:00
Kris Rousey 594f514843 Changing a few remaining instances of passing pods around by value. I 
avoided changing api.PodList.Items to []*api.Pod.
 2015-04-20 15:26:45 -07:00
Wojciech Tyczynski 0e5b478b4d ListMinions support for selectors 2015-04-09 14:22:09 +02:00
gmarek 5a11748018 Move Capacity from NodeSpec to NodeStatus 2015-03-26 15:28:48 +01:00
Masahiro Sano 410e11c305 port endpoints to generic etcd 2015-03-25 22:15:19 +09:00
Clayton Coleman 428d2263e5 Graceful deletion of resources 
This commit adds support to core resources to enable deferred deletion
of resources.  Clients may optionally specify a time period after which
resources must be deleted via an object sent with their DELETE. That
object may define an optional grace period in seconds, or allow the
default "preferred" value for a resource to be used. Once the object
is marked as pending deletion, the deletionTimestamp field will be set
and an etcd TTL will be in place.

Clients should assume resources that have deletionTimestamp set will
be deleted at some point in the future.  Other changes will come later
to enable graceful deletion on a per resource basis.
 2015-03-19 15:33:32 -04:00
Prashanth Balasubramanian 6193608e9b Add an rc strategy, start delta validating updates 2015-03-13 13:09:03 -07:00
Salvatore Dario Minonne 925fa6baf8 Adding fields selector 2015-03-10 22:13:10 +01:00
Brendan Burns 7c654a3d1b Expand test coverage in master, kubectl/cmd/util, pkg/registry/resourcequota, and api/rest. 2015-03-07 15:24:39 -08:00
Brendan Burns 2700871b04 Merge pull request #5013 from smarterclayton/misc_fixup 
Small cleanups to a number of client behaviors
 2015-03-06 15:41:34 +01:00
Mike Danese 5cbf89c060 plumb through changes on ReplicationController and Service 2015-03-04 10:38:42 -08:00
Clayton Coleman a52b0f2619 Switch List/Watch to ListPredicate/WatchPredicate 2015-03-04 10:48:05 -05:00
Paul Morie a131a5e491 Add test for secret RESTStorage 2015-02-18 13:02:22 -05:00
Clayton Coleman 23d199ded9 Make generic etcd more powerful and return operations from etcd 
When we complete an operation, etcd usually provides a response
object.  Return that object up instead of querying etcd twice.
 2015-02-13 13:11:33 -05:00
Brendan Burns 783e1de5bb Add some locking to prevent go data race logic from tripping. 2015-02-05 17:27:58 -08:00
Clayton Coleman 1588970ec4 Turn 409 into 500 Try Again Later when using generateName 
If a client says they want the name to be generated, a 409 is
not appropriate (since they didn't specify a name). Instead, we
should return the next most appropriate error, which is a 5xx
error indicating the request failed but the client *should* try
again.  Since there is no 5xx error that exactly fits this purpose,
use 500 with StatusReasonTryAgainLater set.

This commit does not implement client retry on TryAgainLater, but
clients should retry up to a certain number of times.
 2015-02-02 14:46:25 -05:00
Daniel Smith 2bee4ac572 Merge pull request #3381 from a-robinson/events2 
Improve client recording of events such that clients are
 2015-01-14 15:21:16 -08:00
Paul Morie fd834ae84d Pods should see services only from their own ns 2015-01-14 17:06:36 -05:00
Alex Robinson be6b1cf0e2 Push the full channel logic into the implementation of the broadcaster 
in watch/mux.go rather than being in the client event recording code.
 2015-01-13 02:22:30 +00:00
Clayton Coleman a0d711816d HealthyRegistry should only decorate minions, not omit them 
Fixes #3098
 2015-01-07 11:09:41 -05:00
Deyuan Deng 45bfcb451b Enable watch on node changes. 2014-12-16 08:12:31 +08:00
Daniel Smith 272bfc97ed Merge pull request #2788 from smarterclayton/roundtrip_node_nodelist 
Rename Minions -> Nodes internally
 2014-12-10 11:12:15 -08:00
Clayton Coleman 19379b5a38 Internal rename api.Minion -> api.Node 2014-12-10 12:08:18 -05:00
Deyuan Deng 76552423f9 Move watch filter into storage level 2014-12-04 20:15:27 -05:00
Eric Tune 3f285269cc Rename watch.Mux -> watch.Broadcaster 
A few reasons:
- Mux is already widely used in the codebase to refer to a http handler mux.
- Original meaning of Mux was something which sent a chose one of several inputs to
  and output.  This sends one output to all outputs.  Broadcast captures that idea
  better.
- Aligns with similar class config.Broadcaster (see #2747)
 2014-12-04 00:30:51 -08:00
Clayton Coleman 156000ef6d Move the internal minion representation to match v1beta3 
Moves to 'Spec' and 'Status' internally and removes duplicate
fields.  Moves Capacity into Spec and drops use of NodeResources
 2014-11-20 20:53:08 -05:00
Deyuan Deng cc310e0e71 Support node label update. 2014-11-17 13:42:31 -05:00
Daniel Smith 178d0af795 Fix watch for events; add test for interface implementation so it won't break again. 2014-11-12 15:22:08 -08:00
Tim Hockin bb5ca0f91b Respect a requested PortalIP if possible 2014-10-27 14:03:52 -07:00
Clayton Coleman 7550c146dc Replace struct initializers for TypeMeta with ObjectMeta 2014-10-24 11:22:21 -04:00
Clayton Coleman bb77a5d15f Rename ID -> Name 2014-10-22 15:00:26 -04:00
Tim Hockin e907011111 Core support for ip-per-service 2014-10-16 08:36:47 -07:00
Deyuan Deng ec46e94dc2 Create MinionController to sync minions from cloudprovider (pkg cloudprovider/controller). 2014-10-14 18:45:25 -04:00
Daniel Smith 9a9362e896 Add generic registry object so we can stop rewriting this code 2014-10-10 15:46:49 -07:00
Deyuan Deng 15e5b876b8 Change ContainsMinion to GetMinion 2014-10-08 23:41:15 -04:00
Clayton Coleman 82bcdd3b3b Make ResourceVersion a string internally instead of uint64 
Allows us to define different watch versioning regimes in the future
as well as to encode information with the resource version.

This changes /watch/resources?resourceVersion=3 to start the watch at
4 instead of 3, which means clients can read a resource version and
then send it back to the server. Clients should no longer do math on
resource versions.
 2014-10-07 19:00:26 -04:00
Deyuan Deng c3d9197a4b Rename InsertMinion to CreateMinion. 2014-10-07 16:22:35 -04:00
Deyuan Deng 4a35325f29 Use etcd as backend for minion registry. 2014-10-07 16:22:35 -04:00
Clayton Coleman d3e51a0f24 Rename JSONBase -> TypeMeta in preparation for v1beta3 
Will make subsequent refactor much easier
 2014-10-07 11:12:16 -04:00
Deyuan Deng 7bae027604 Fix bug in minion registrytest. 2014-10-05 15:55:26 -04:00
Brendan Burns 1551b48347 Add a resource fit scheduler predicate. Set sensible defaults. 2014-10-03 15:09:59 -07:00
derekwaynecarr b7fcc7d3ec Add ctx to registry test cases 2014-09-26 15:19:22 -04:00
derekwaynecarr ee19ba186d Update to use api.Context 2014-09-26 11:50:34 -04:00
derekwaynecarr be85ad7a3d Add context object in test cases flows 2014-09-26 11:50:34 -04:00
Deyuan Deng 0c8d9ba870 Fix minion registry test insert. 2014-09-18 23:04:10 -04:00
Daniel Smith 4e9cf2ccb4 Add field selector to List, implement for pods. 2014-09-16 16:17:16 -07:00
Daniel Smith fc09f988b4 Make tests pass again 2014-09-07 22:26:42 -07:00
Clayton Coleman 01e668187c Services and Endpoints weren't properly sync'ing 
They need incremental changes and a resync on start.
 2014-09-03 16:04:55 -04:00
Vojtech Vitek (V-Teq) 42c656501c Add several missing pkg Godoc files 
Signed-off-by: Vojtech Vitek (V-Teq) <vvitek@redhat.com>
 2014-09-02 13:41:11 +02:00
Clayton Coleman bafc422ac0 Add the resource version to api.*List items from etcd 
Allows clients to watch more easily (can invoke Get, then
Watch).
 2014-08-28 23:17:00 -04:00
Daniel Smith 4de254444e Add simple service redirection 2014-08-27 20:51:41 -07:00
Clayton Coleman 083d81b6d7 Expose REST resource for endpoints and watch on services/endpoints 
Will allow kube-proxies to listen on endpoints.
 2014-08-27 15:49:01 -04:00
Daniel Smith 72b35816cd Need to remove pods that change labels. 2014-08-25 15:56:54 -07:00
Daniel Smith 29e9e13188 Remove synchronous assignPod call from create pod 2014-08-25 11:59:00 -07:00
Deyuan Deng c104551025 Improve pkg/registry/service code coverage. 2014-08-19 20:13:00 -04:00
Daniel Smith 4b2867fd8a Standardize watch usage in registry/storage objects. Fix up extremely confusing pod test object. 2014-08-16 19:07:41 -07:00
Daniel Smith e40e5b53a7 Add Watch to /pods. 2014-08-16 19:07:29 -07:00
Daniel Smith 26e2256178 Remove unused and not completely correct code 2014-08-15 18:20:37 -07:00
Clayton Coleman 231ed95a61 Typo in name 2014-08-13 14:36:23 -04:00
Kelsey Hightower c21a0ca39f Breakup the registry package into separate packages. 
Currently all registry implementations live in a single package,
which makes it bit harder to maintain. The different registry
implementations do not follow the same coding style and naming
conventions, which makes the code harder to read.

Breakup the registry package into smaller packages based on
the registry implementation. Refactor the registry packages
to follow a similar coding style and naming convention.

This patch does not introduce any changes in behavior.
 2014-08-11 20:58:09 -07:00