upstream/kubernetes
1
0
Fork 0
mirror of https://github.com/kubernetes/kubernetes.git synced 2026-06-08 16:30:57 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions
138342 commits 61 branches 1227 tags 2.8 GiB
Commit graph

138342 commits

This branch
This branch
All branches
Author SHA1 Message Date
Kubernetes Prow Robot
6aaa84c0a0
Merge pull request #139549 from serathius/watchcache-benchmark-watchlatency 
Implement watch event delivery latency tracking in write throughput benchmark
 2026-06-09 01:31:54 +05:30
Kubernetes Prow Robot
de7f0f85c8
Merge pull request #139533 from serathius/watchcache-snapshot-lock 
List snapshot data outside of lock
 2026-06-09 01:31:47 +05:30
Kubernetes Prow Robot
1fd96eb4bb
Merge pull request #137694 from Jefftree/client-go-throttle-err 
client-go: return context error directly instead of wrapping as rate limiter error
 2026-06-08 23:33:47 +05:30
Marek Siarkowicz
776bd20c20 Implement watch event delivery latency tracking in write throughput benchmark 
This change replaces the seconds-delay metric with P99 watch-latency, which
averages event delivery latency across all events during the parallel run.
This significantly reduces measurement variance under high lock contention:
- Previous approach (seconds-delay): ±37,839% variance
- New approach (P99 watch-latency): ±327% variance

Signed-off-by: Marek Siarkowicz <siarkowicz@google.com>
 2026-06-08 19:52:06 +02:00
Kubernetes Prow Robot
2f5e22e8bb
Merge pull request #139536 from serathius/watchcache-benchmark-exactrv 
Implement Exact RV LIST background load in write throughput benchmark
 2026-06-08 21:09:47 +05:30
Kubernetes Prow Robot
9560c473da
Merge pull request #139245 from dom4ha/fix-kubelet-windows-test-cleanup 
Fix Windows file locking race condition in Kubelet tests
 2026-06-08 21:09:40 +05:30
Marek Siarkowicz
44500eca4f List snapshot data outside of lock 
This optimizes the exact resource version list logic (StoreList/.../RV=Exact)
by extracting the snapshot and performing the B-Tree traversal outside of
the watch cache lock.

The benchstat evidence below is limited to the `RV=Exact` scenarios, which
are the only paths executing the modified code.

```text

goos: linux
goarch: amd64
pkg: k8s.io/apiserver/pkg/storage/cacher
cpu: AMD Ryzen Threadripper PRO 3945WX 12-Cores
                                                                                                                                         │ before_exact_fixed.txt │        after_exact_fixed.txt         │
                                                                                                                                         │         sec/op         │    sec/op      vs base               │
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=DeleteCreate/Parallelism=25/Background=ListerExactRV/UseIndex=false-24           218.4µ ±   37%   109.9µ ±  46%  -49.68% (p=0.004 n=6)
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=DeleteCreate/Parallelism=25/Background=ListerExactRV/UseIndex=true-24            215.2µ ±   48%   218.9µ ±  47%        ~ (p=0.937 n=6)
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=Patch/Parallelism=25/Background=ListerExactRV/UseIndex=false-24                  50.88µ ± 1960%   40.65µ ± 157%        ~ (p=0.818 n=6)
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=Patch/Parallelism=25/Background=ListerExactRV/UseIndex=true-24                   37.41µ ±   32%   25.05µ ±  58%        ~ (p=0.180 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=true/RV=Exact/Scope=Cluster/Paginate=0-24                                                   30.33m ±   24%   24.79m ±  35%        ~ (p=0.240 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=true/RV=Exact/Scope=Cluster/Paginate=1000-24                                                 2.254 ±    9%    2.194 ±  12%        ~ (p=0.589 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=true/RV=Exact/Scope=Node/Paginate=0-24                                                      7.960m ±   15%   8.118m ±  15%        ~ (p=0.589 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=true/RV=Exact/Scope=Namespace/Paginate=0-24                                                 453.6µ ±    5%   429.1µ ±  18%   -5.40% (p=0.041 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=true/RV=Exact/Scope=Namespace/Paginate=1000-24                                              497.3µ ±   15%   514.2µ ±  13%        ~ (p=1.000 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=false/RV=Exact/Scope=Cluster/Paginate=0-24                                                  27.38m ±   10%   25.53m ±  39%        ~ (p=0.485 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=false/RV=Exact/Scope=Cluster/Paginate=1000-24                                                2.355 ±   16%    2.057 ±  18%        ~ (p=0.240 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=false/RV=Exact/Scope=Node/Paginate=0-24                                                     25.19m ±   47%   24.64m ±  27%        ~ (p=0.699 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=false/RV=Exact/Scope=Namespace/Paginate=0-24                                                422.5µ ±   20%   423.2µ ±  11%        ~ (p=0.699 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=false/RV=Exact/Scope=Namespace/Paginate=1000-24                                             499.4µ ±   11%   530.0µ ±  26%        ~ (p=0.132 n=6)
geomean                                                                                                                                            2.949m           2.613m         -11.39%

                                                                                                                                         │ before_exact_fixed.txt │       after_exact_fixed.txt        │
                                                                                                                                         │      list-calls/s      │ list-calls/s  vs base              │
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=DeleteCreate/Parallelism=25/Background=ListerExactRV/UseIndex=false-24            8.180 ±  102%    9.119 ± 38%       ~ (p=0.485 n=6)
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=DeleteCreate/Parallelism=25/Background=ListerExactRV/UseIndex=true-24             8.334 ±   67%    7.505 ± 52%       ~ (p=0.699 n=6)
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=Patch/Parallelism=25/Background=ListerExactRV/UseIndex=false-24                   8.194 ± 2200%    8.233 ± 27%       ~ (p=0.937 n=6)
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=Patch/Parallelism=25/Background=ListerExactRV/UseIndex=true-24                    9.043 ±   31%    9.500 ± 22%       ~ (p=0.818 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=true/RV=Exact/Scope=Cluster/Paginate=0-24                                                    32.99 ±   31%    40.41 ± 44%       ~ (p=0.240 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=true/RV=Exact/Scope=Cluster/Paginate=1000-24                                                 66.54 ±    9%    68.38 ± 14%       ~ (p=0.589 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=true/RV=Exact/Scope=Node/Paginate=0-24                                                       125.7 ±   13%    123.2 ± 17%       ~ (p=0.589 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=true/RV=Exact/Scope=Namespace/Paginate=0-24                                                 2.205k ±    5%   2.331k ± 23%  +5.71% (p=0.041 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=true/RV=Exact/Scope=Namespace/Paginate=1000-24                                              6.034k ±   13%   5.835k ± 15%       ~ (p=1.000 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=false/RV=Exact/Scope=Cluster/Paginate=0-24                                                   36.52 ±   11%    39.28 ± 28%       ~ (p=0.485 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=false/RV=Exact/Scope=Cluster/Paginate=1000-24                                                63.70 ±   19%    72.92 ± 15%       ~ (p=0.240 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=false/RV=Exact/Scope=Node/Paginate=0-24                                                      39.73 ±   32%    40.83 ± 22%       ~ (p=0.699 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=false/RV=Exact/Scope=Namespace/Paginate=0-24                                                2.366k ±   25%   2.363k ± 10%       ~ (p=0.699 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=false/RV=Exact/Scope=Namespace/Paginate=1000-24                                             6.007k ±   12%   5.665k ± 35%       ~ (p=0.132 n=6)
geomean                                                                                                                                             106.6            110.2        +3.38%

                                                                                                                                         │ before_exact_fixed.txt │       after_exact_fixed.txt        │
                                                                                                                                         │      list-objs/s       │ list-objs/s   vs base              │
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=DeleteCreate/Parallelism=25/Background=ListerExactRV/UseIndex=false-24           1.225M ±  102%   1.366M ± 38%       ~ (p=0.394 n=6)
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=DeleteCreate/Parallelism=25/Background=ListerExactRV/UseIndex=true-24             275.0 ±   67%    216.4 ± 53%       ~ (p=0.485 n=6)
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=Patch/Parallelism=25/Background=ListerExactRV/UseIndex=false-24                  1.229M ± 2200%   1.235M ± 27%       ~ (p=0.937 n=6)
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=Patch/Parallelism=25/Background=ListerExactRV/UseIndex=true-24                    298.4 ±   31%    275.5 ± 22%       ~ (p=0.180 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=true/RV=Exact/Scope=Cluster/Paginate=0-24                                                   4.947M ±   31%   6.059M ± 44%       ~ (p=0.240 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=true/RV=Exact/Scope=Cluster/Paginate=1000-24                                                66.50k ±    9%   68.34k ± 14%       ~ (p=0.589 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=true/RV=Exact/Scope=Node/Paginate=0-24                                                      3.764k ±   13%   3.735k ± 17%       ~ (p=0.589 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=true/RV=Exact/Scope=Namespace/Paginate=0-24                                                 6.614M ±    5%   6.992M ± 23%  +5.72% (p=0.041 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=true/RV=Exact/Scope=Namespace/Paginate=1000-24                                              6.032M ±   13%   5.835M ± 15%       ~ (p=1.000 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=false/RV=Exact/Scope=Cluster/Paginate=0-24                                                  5.476M ±   11%   5.890M ± 28%       ~ (p=0.485 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=false/RV=Exact/Scope=Cluster/Paginate=1000-24                                               63.66k ±   19%   72.88k ± 15%       ~ (p=0.240 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=false/RV=Exact/Scope=Node/Paginate=0-24                                                     5.958M ±   32%   6.121M ± 22%       ~ (p=0.699 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=false/RV=Exact/Scope=Namespace/Paginate=0-24                                                7.099M ±   25%   7.088M ± 10%       ~ (p=0.699 n=6)
StoreList/Namespaces=50/Pods=150000/Nodes=5000/Indexed=false/RV=Exact/Scope=Namespace/Paginate=1000-24                                             6.006M ±   12%   5.664M ± 35%       ~ (p=0.132 n=6)
geomean                                                                                                                                            356.7k           362.2k        +1.54%

                                                                                                                                         │ before_exact_fixed.txt │         after_exact_fixed.txt          │
                                                                                                                                         │     seconds-delay      │  seconds-delay   vs base               │
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=DeleteCreate/Parallelism=25/Background=ListerExactRV/UseIndex=false-24          616.95m ±  157%    30.28m ± 1951%  -95.09% (p=0.026 n=6)
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=DeleteCreate/Parallelism=25/Background=ListerExactRV/UseIndex=true-24            649.4m ±  164%   1287.5m ±   85%        ~ (p=0.394 n=6)
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=Patch/Parallelism=25/Background=ListerExactRV/UseIndex=false-24                  30.70m ± 1818%    55.93m ± 1053%        ~ (p=0.589 n=6)
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=Patch/Parallelism=25/Background=ListerExactRV/UseIndex=true-24                   27.02m ±   38%    27.83m ± 1378%        ~ (p=0.589 n=6)
geomean                                                                                                                                            135.0m            88.26m          -34.64%

                                                                                                                                         │ before_exact_fixed.txt │        after_exact_fixed.txt        │
                                                                                                                                         │        writes/s        │   writes/s    vs base               │
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=DeleteCreate/Parallelism=25/Background=ListerExactRV/UseIndex=false-24             19.92k ± 41%   22.37k ± 16%        ~ (p=0.394 n=6)
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=DeleteCreate/Parallelism=25/Background=ListerExactRV/UseIndex=true-24              22.23k ± 22%   23.36k ± 27%        ~ (p=0.485 n=6)
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=Patch/Parallelism=25/Background=ListerExactRV/UseIndex=false-24                    25.92k ± 59%   27.93k ± 44%        ~ (p=1.000 n=6)
StoreWriteThroughput/Namespaces=50/Pods=150000/Nodes=5000/Traffic=Patch/Parallelism=25/Background=ListerExactRV/UseIndex=true-24                     32.92k ± 55%   48.04k ± 36%        ~ (p=0.132 n=6)
geomean                                                                                                                                              24.79k         28.93k        +16.71%

Signed-off-by: Marek Siarkowicz <siarkowicz@google.com>
 2026-06-08 14:27:55 +02:00
Kubernetes Prow Robot
42d9ed9c47
Merge pull request #139374 from brejman/placement-feasible-init-fast-fail 
Fail if PlacementFeasible is not implemented by GangScheduling
 2026-06-08 17:23:45 +05:30
Kubernetes Prow Robot
a006a3c610
Merge pull request #139547 from HirazawaUi/fix-flaky-test 
Fix flaking e2e_node tests
 2026-06-07 23:54:30 +05:30
HirazawaUi
e79d1a4271 Fix flaking e2e_node tests 2026-06-07 15:11:57 +08:00
Kubernetes Prow Robot
643e407efe
Merge pull request #139532 from serathius/watchcache-waituntilfresh-cleanup 
Cleanup if statements by making waitUntilFreshLocked to handle sending progress notifies if consistent read is supported
 2026-06-06 22:14:27 +05:30
Marek Siarkowicz
1bc56741eb Implement Exact RV LIST background load in write throughput benchmark 
This change extends the write throughput benchmark background load options with
ListerExactRV, which queries the watch cache with exact resource version matching
to avoid database consistency read roundtrips.

Comparing Lister vs ListerExactRV under full dimensions (150k pods):

                                                │ new_lister_full.bench │

                                                │        sec/op         │
StoreWriteThroughput/.../Background=Lister/...                 24.06µ ± 323%
StoreWriteThroughput/.../Background=ListerExactRV/...          17.81µ ±   8%
geomean                                                        20.70µ

                                                │     list-calls/s      │
StoreWriteThroughput/.../Background=Lister/...                  72.41 ±  79%
StoreWriteThroughput/.../Background=ListerExactRV/...           9.468 ±  11%
geomean                                                         26.19

                                                │      list-objs/s      │
StoreWriteThroughput/.../Background=Lister/...                 2.462k ±  79%
StoreWriteThroughput/.../Background=ListerExactRV/...          321.9  ±  11%
geomean                                                        890.2

                                                │       writes/s        │
StoreWriteThroughput/.../Background=Lister/...                 54.78k ±  45%
StoreWriteThroughput/.../Background=ListerExactRV/...          63.67k ±   6%
geomean                                                        59.06k

By querying historical snapshots via ListerExactRV, the write throughput variance
drops from ±45% to ±6%, and the execution time variance drops from ±323% to ±8%.

Note on NotOlderThan: Evaluated switching Lister to NotOlderThan (stale list), but
due to high lock contention on the live cache store in the unoptimized cacher, the
variance profile was identical to consistent listing (±79%). We have therefore
retained the default Lister as consistent read.

Signed-off-by: Marek Siarkowicz <siarkowicz@google.com>
 2026-06-06 15:29:43 +02:00
Kubernetes Prow Robot
6070d2808d
Merge pull request #139516 from Vinayak9769/fix-nftables-comment-truncation 
proxy/nftables: Truncate service comments to prevent length limit violations
 2026-06-06 17:40:37 +05:30
Marek Siarkowicz
d9c96f34c5 Cleanup if statements by making waitUntilFreshLocked to handle sending progress notifies if consistent read is supported 2026-06-06 11:54:47 +02:00
Kubernetes Prow Robot
e053c57914
Merge pull request #139527 from Jefftree/range-stream-benchmarking 
apiserver: exercise RangeStream on/off in BenchmarkCacherInit
 2026-06-06 15:16:30 +05:30
Kubernetes Prow Robot
6b364b1811
Merge pull request #137988 from Jefftree/flake-fix 
leaderelection: retry release on conflict from inflight renew
 2026-06-06 04:40:40 +05:30
Kubernetes Prow Robot
546c46f0d8
Merge pull request #139006 from rzlink/fix/e2e-common-race-conditions 
test/e2e/common/node: poll for eventually-consistent state to reduce flakiness
 2026-06-06 03:36:32 +05:30
Jefftree
5d3ddc2119 apiserver: exercise RangeStream on/off in BenchmarkCacherInit 2026-06-05 16:44:54 -04:00
Vinayak Mohanty
4d306fc68e
fix: truncate service comments in nftables to prevent length limit violations 
refactor: rename svcPortNameString to svcPortComment and update test validation in nftables proxier
 2026-06-06 01:49:08 +05:30
Kubernetes Prow Robot
6ebbd2fc51
Merge pull request #139495 from serathius/watchcache-locking 
Reduce times lock is aquired in watch cache during reads
 2026-06-06 00:51:59 +05:30
Kubernetes Prow Robot
1b7d495d61
Merge pull request #136915 from Jefftree/range-stream 
KEP 5966: Implement RangeStream for watch cache
 2026-06-05 22:39:55 +05:30
Kubernetes Prow Robot
b7de3e336d
Merge pull request #139524 from serathius/watchcache-benchmark-conflict 
Reduce conflicts in benchmark between PATCH requests by making them sequential
 2026-06-05 20:57:54 +05:30
Jefftree
33d1d2e656 apiserver: inline isUnimplementedErr and clarify sync fallback flow 2026-06-05 11:07:21 -04:00
Jefftree
dbbc4dc176 apiserver: interpret RangeStream errors in syncStreamRecursive 2026-06-05 10:38:38 -04:00
Marek Siarkowicz
6736a03587 Minimize conflicts between PATCH requests by making them sequential 2026-06-05 16:02:05 +02:00
Kubernetes Prow Robot
7c7605a853
Merge pull request #139330 from brejman/fix-stuck-preemption-followup 
Unset WasFlushedFromUnschedulable for gated pods
 2026-06-05 18:33:55 +05:30
Kubernetes Prow Robot
f92c74d803
Merge pull request #139499 from serathius/watchcache-list-benchmark 
Improve concurrent store list benchmarking
 2026-06-04 23:05:53 +05:30
Marek Siarkowicz
f4555eaa7e Reduce times lock is aquired in watch cache during reads 2026-06-04 14:23:44 +02:00
Kubernetes Prow Robot
9fa4c1cfa3
Merge pull request #139397 from adrianmoisey/ipvs-feature-gate 
KEP-5495: Add featuregate for IPVS
 2026-06-04 17:47:48 +05:30
Marek Siarkowicz
216ab1b334 Improve concurrent store list benchmarking 2026-06-04 12:22:41 +02:00
Kubernetes Prow Robot
0d31ea790c
Merge pull request #139480 from lalitc375/fix-wrong-marking 
Fix wrong marking of errors
 2026-06-04 04:31:46 +05:30
Kubernetes Prow Robot
e2f5ec83d8
Merge pull request #139479 from BenTheElder/go1.26.4 
upgrade go to 1.26.4
 2026-06-04 02:51:48 +05:30
Lalit Chauhan
0c84f84ec8 Fix wrong marking of errors 2026-06-03 19:33:53 +00:00
Benjamin Elder
c90b6dbecb upgrade go to 1.26.4 2026-06-03 12:07:03 -07:00
Adrian Moisey
f7265100cb
KEP-5495: Add featuregate for IPVS 2026-06-03 21:04:16 +02:00
Kubernetes Prow Robot
fbcbb81625
Merge pull request #139467 from serathius/watchcache-btree-cleanup 
Cleanup ordered listing after removal of btree feature gate
 2026-06-04 00:05:46 +05:30
Kubernetes Prow Robot
862d42080f
Merge pull request #139466 from soltysh/maxunavailable_beta3 
KEP-961: maxUnavailable feature on by default
 2026-06-03 22:09:55 +05:30
Kubernetes Prow Robot
a0afe51e25
Merge pull request #139129 from pohly/e2e-node-update-local 
E2E node: enable using release archives for periodic jobs, simplified
 2026-06-03 22:09:47 +05:30
Marek Siarkowicz
94a03e35c0 Cleanup ordered listing after removal of btree feature gate 2026-06-03 18:36:08 +02:00
Kubernetes Prow Robot
7a1385a332
Merge pull request #139402 from aman4433/svm-allow-notfound-patches 
test/integration: fix flaky TestStorageVersionMigrationDuringChaos by extracting audit validation helper
 2026-06-03 20:49:59 +05:30
Kubernetes Prow Robot
40dab40837
Merge pull request #138856 from RomanBednar/handle-corrupted-subpath-mount 
Handle corrupted subpath mount points during container restart
 2026-06-03 20:49:51 +05:30
Maciej Szulik
29915cc905
KEP-961: maxUnavailable feature on by default 
Signed-off-by: Maciej Szulik <soltysh@gmail.com>
 2026-06-03 15:44:57 +02:00
Jefftree
3ea54dafff apiserver: address RangeStream review feedback 2026-06-03 09:21:24 -04:00
Kubernetes Prow Robot
e0c1c5bd3d
Merge pull request #138488 from yliaog/master 
DRA Extended Resource: promote to GA in 1.37
 2026-06-03 18:39:48 +05:30
Kubernetes Prow Robot
a94f8bfc3e
Merge pull request #139280 from antekjb/supportNNNinPodgrupPreamption 
Make podgroup preemption integration test to check NNN info without setting clearingNominatedNodeNameAfterBinding to flase
 2026-06-03 17:29:46 +05:30
yliao
523e3c8500 removed the test cases that disable DRAExtendedResource feature 2026-06-03 11:17:09 +00:00
Patrick Ohly
de2d13b27e e2e_node: support pre-built binaries 
This is not usable through "make test-e2e-node", which (while feasible) would
be a bit pointless because the Kubernetes source could would still be needed
for the make rules.

Instead, "kubetest2 noop -test=node" gets extended to invoke `e2e_node.test
remote` with flags that tell e2e_node.test where to find the binaries and
flags that were provided by the caller of kubetest2.
 2026-06-03 10:32:48 +02:00
Kubernetes Prow Robot
48d5bff656
Merge pull request #139452 from serathius/watchcache-init-ga 
Graduate WatchCacheInitializationPostStartHook to GA
 2026-06-03 13:47:46 +05:30
Patrick Ohly
2d574790a6 e2e_node: fix log output 
fmt.Printf lacked the trailing newline and is inconsistent with other output,
which uses klog.
 2026-06-03 08:34:56 +02:00
Patrick Ohly
6ba4d21765 e2e_node: multiplex different commands in e2e_node.test 
The additional commands (mounter, gcp-credentials-provider) are needed for E2E
node testing. This change makes e2e_node.test entirely self-contained.

Copying the commands' code into separate packages is temporary and only done to
avoid touching them while it is still unclear whether this approach will work
out.

Besides avoiding changes to the build rules, bundling the functionality also has a
slight size advantage: the size of e2e_node.test increases by 10KB, whereas
the other two separate commands would add 10MB.
 2026-06-03 08:34:56 +02:00