Sok3t/descheduler
1
0
Fork 0
mirror of https://github.com/kubernetes-sigs/descheduler.git synced 2026-01-26 13:29:11 +01:00
Code Packages Releases Activity
1,032 Commits 21 Branches 83 Tags
224e2b078f62e3537d32caec7733f0918d48c248
Commit Graph

375 Commits

Author SHA1 Message Date
Julian Lawrence
224e2b078f updated logs to help with debugging 2022-02-14 18:27:53 -08:00
Julian Lawrence
1a98a566b3 adding cancelation from sigint sigterm 2022-01-25 00:10:09 -08:00
Jan Chaloupka
203388ff1a Evictor: report successful eviction independently of the dry-mode 
Dry mode currently does not report metrics when the eviction succeeds
 2022-01-20 21:23:19 +01:00
Jan Chaloupka
901a16ecbc Do not collect the metrics when the metrics server is not enabled 2022-01-20 17:04:15 +01:00
Jan Chaloupka
e0f086ff85 Use a fake client when evicting pods by individual strategies to accumulate the evictions 
Currently, when the descheduler is running with the --dry-run on, no strategy actually
evicts a pod so every strategy always starts with a complete list of
pods. E.g. when the PodLifeTime strategy evicts few pods, the RemoveDuplicatePods
strategy still takes into account even the pods eliminated by the PodLifeTime
strategy. Which does not correspond to the real case scenarios as the
same pod can be evicted multiple times. Instead, use a fake client and
evict/delete the pods from its cache so the strategies evict each pod
at most once as it would be normally done in a real cluster.
 2022-01-20 17:04:05 +01:00
Kubernetes Prow Robot
551eced42a Merge pull request #688 from babygoat/evict-failed-without-ownerrefs 
feat: support eviction of failed bare pods
 2022-01-11 12:31:15 -08:00
Julian Lawrence
77cb406052 updated until -> sliding until 2022-01-06 12:55:10 -08:00
babygoat
1529180d70 feat: support eviction of failed bare pods 
This patch adds the policy(evictFailedBarePods) to allow the failed
pods without ownerReferences to be evicted. For backward compatibility,
disable the policy by default. Address #644.
 2022-01-06 01:07:41 +08:00
Jan Chaloupka
8462cf56d7 run ./hack/update-* scripts 2022-01-04 09:37:01 +01:00
Erik Jankovič
2020642b6f chore: add pod.Status.Reason to the list of reasons 
Signed-off-by: Erik Jankovič <erik.jankovic@gmail.com>
 2021-12-17 18:37:53 +01:00
Kubernetes Prow Robot
1e5165ba9f Merge pull request #670 from autumn0207/improve_pod_eviction_metrics 
Add node name label to the counter metric for evicted pods
 2021-12-16 01:49:18 -08:00
autumn0207
8e74f8bd77 improve pod eviction metrics 2021-12-16 17:06:22 +08:00
Kubernetes Prow Robot
2424928019 Merge pull request #667 from damemi/1.23-rc.0 
bump: k8s to 1.23
 2021-12-15 06:56:20 -08:00
Garrybest
cac3b9185b reform all test files 
Signed-off-by: Garrybest <garrybest@foxmail.com>
 2021-12-11 19:43:16 +08:00
Mike Dame
4fa7bf978c run hack/update-generated-deep-copies.sh 2021-12-10 10:02:39 -05:00
Garrybest
0ff8ecb41e reform all strategies by using getPodsAssignedToNode 
Signed-off-by: Garrybest <garrybest@foxmail.com>
 2021-12-10 19:28:51 +08:00
Garrybest
08ed129a07 reform ListPodsOnANode by using pod informer and indexer 
Signed-off-by: Garrybest <garrybest@foxmail.com>
 2021-12-10 19:25:20 +08:00
Kubernetes Prow Robot
82201d0e48 Add maxNoOfPodsToEvictPerNamespace policy 2021-12-03 10:58:37 +08:00
Jan Chaloupka
73a7adf572 Drop deprecated flags 2021-11-29 17:12:59 +01:00
Jan Chaloupka
4603182320 Bump go version in go.mod to go1.17 2021-11-29 16:49:35 +01:00
Jane Liu L
c7524705b3 Ignore Pods With Deletion Timestamp 2021-11-10 09:32:11 +08:00
Kubernetes Prow Robot
3be910c238 Merge pull request #621 from uthark/oatamanenko/deleted 
Ignore pods being deleted
 2021-09-30 21:21:22 -07:00
Amir Alavi
f7c26ef41f e2e tests for RemoveFailedPods strategy 
Fix priority class default
 2021-09-26 20:39:32 -04:00
Amir Alavi
6eb37ce079 RemoveFailedPods: guard against nil descheduler strategy (e.g. in case of default that loads all strategies) 2021-09-20 11:20:54 -04:00
kaiyuechen
d1d9ea0c48 fix duplicate code in node_affinity_test.go 2021-09-16 10:39:52 +08:00
Oleg Atamanenko
4448d9c670 Ignore pods being deleted 2021-09-15 00:05:51 -07:00
Amir Alavi
0e0e688fe8 Introduce RemoveFailedPods strategy 2021-08-30 14:17:52 -04:00
Daniel Klockenkämper
a6b34c1130 nodeutilization strategy: "numberOfNodes" should work as documented 2021-08-17 13:29:07 +00:00
Amir Alavi
b0e5d64bd7 TopologySpreadConstraint: advance above avg index when at ideal average when balancing domains 2021-07-13 22:55:10 -04:00
Jan Chaloupka
d998d82357 HighNodeUtilization: add NodeFit feature 2021-06-08 16:59:43 +02:00
Kubernetes Prow Robot
3843a2d5d1 Merge pull request #550 from hanumanthan/highnodeutilisation 
Highnodeutilization strategy
 2021-06-08 01:59:12 -07:00
Hanu
4cd1e66ef3 Adding highnodeutilization strategy 2021-06-06 18:01:42 +08:00
Hanu
2f18864fa5 Refractor - Modify the common functions to be used by high utilisation 2021-06-06 18:00:43 +08:00
Hanu
6e71068f4f Refractoring lownodeutilization - extracting common functions 2021-06-06 18:00:29 +08:00
Amir Alavi
012ca2398f Filter pods by labelSelector during eviction for TopologySpreadConstraint strategy 2021-06-01 15:42:23 -04:00
RyanDevlin
41d46d0d3b Working nodeFit feature 2021-05-24 09:03:38 -04:00
Kubernetes Prow Robot
3b9d3d9719 Merge pull request #563 from ingvagabund/removeduplicates-take-taints-into-account-for-node-count 
RemoveDuplicates: take node taints, node affinity and node selector into account when computing a number of feasible nodes for the average occurence of pods per node
 2021-05-21 05:38:46 -07:00
Jan Chaloupka
5396282e3d RemoveDuplicates: take node taints, node affinity and node selector into account when computing a number of feasible nodes for the average occurence of pods per node 
Nodes with taints which are not tolerated by evicted pods will never run the
pods. The same holds for node affinity and node selector.
So increase the number of pods per feasible nodes to decrease the
number of evicted pods.
 2021-05-18 16:13:24 +02:00
BinacsLee
fe8e17f72c fix staticcheck failure for pkg/descheduler/descheduler_test.go 2021-05-17 23:07:12 +08:00
Amir Alavi
24c0ca2ef9 Take node's taints into consideration when balancing domains 2021-05-14 15:23:58 -04:00
Mike Dame
fc83c13166 Add test cases for soft constraints/multi constraints 2021-05-12 08:36:07 -04:00
Jan Chaloupka
4edbecc85d Define NodeSelectorsEqual predicate 2021-05-09 18:08:32 +02:00
Jan Chaloupka
54f67266bb Define TolerationsEqual 2021-05-09 18:08:27 +02:00
Sean Malloy
6bde95c9a1 Use Structured Logging For Unknown Strategy Log Message 
Always use structured logging. Therefore update klog.Errorf() to instead
use klog.ErrorS().

Here is an example of the new log message.

E0428 23:58:57.048912 586 descheduler.go:145] "skipping strategy" err="unknown strategy name" strategy=ASDFPodLifeTime
 2021-04-29 00:00:07 -05:00
Mike Dame
feae158a50 Invert main strategy loop for performance and customizability 2021-04-28 10:36:02 -04:00
BinacsLee
c4afb6bb30 code cleanup: remove check on length 2021-04-25 21:44:20 +08:00
Xiang Liu
a848dac3cf Updating policy api version used in pod evictor 2021-04-13 11:02:01 +08:00
Jan Chaloupka
c239e1199f LNU: improve nodeUsage logging 
To avoid:
```
I0210 11:56:04.137956 3309277 lownodeutilization.go:389] "Updated node usage" updatedUsage={node:0xc000460000 usage:map[cpu:0xc00042b480 memory:0xc00042b4c0 pods:0xc00042b500] allPods:[0xc0004a0000 0xc0004a03e8 0xc0004a07d0 0xc0004a0bb8 0xc0004a0fa0 0xc0004a1388 0xc0004a1770 0xc0004a1b58] lowResourceThreshold:map[cpu:0xc00042b540 memory:0xc00042b580 pods:0xc00042b5c0] highResourceThreshold:map[cpu:0xc00042b600 memory:0xc00042b640 pods:0xc00042b680]}
I0210 11:56:04.138829 3309277 lownodeutilization.go:389] "Updated node usage" updatedUsage={node:0xc000460000 usage:map[cpu:0xc00042b480 memory:0xc00042b4c0 pods:0xc00042b500] allPods:[0xc0004a0000 0xc0004a03e8 0xc0004a07d0 0xc0004a0bb8 0xc0004a0fa0 0xc0004a1388 0xc0004a1770 0xc0004a1b58] lowResourceThreshold:map[cpu:0xc00042b540 memory:0xc00042b580 pods:0xc00042b5c0] highResourceThreshold:map[cpu:0xc00042b600 memory:0xc00042b640 pods:0xc00042b680]}
I0210 11:56:04.139044 3309277 lownodeutilization.go:389] "Updated node usage" updatedUsage={node:0xc000460000 usage:map[cpu:0xc00042b480 memory:0xc00042b4c0 pods:0xc00042b500] allPods:[0xc0004a0000 0xc0004a03e8 0xc0004a07d0 0xc0004a0bb8 0xc0004a0fa0 0xc0004a1388 0xc0004a1770 0xc0004a1b58] lowResourceThreshold:map[cpu:0xc00042b540 memory:0xc00042b580 pods:0xc00042b5c0] highResourceThreshold:map[cpu:0xc00042b600 memory:0xc00042b640 pods:0xc00042b680]}
```
 2021-04-06 09:43:23 +02:00
ZongqiangZhang
81b816d4a4 support extended resources in lownodeutilization 2021-04-02 21:37:51 +08:00
Sean Malloy
af01b675b0 Update Generated Code 
Ran "make gen" using Go 1.16.1. Some changes were merged, but "make gen"
was not run. This fixes the problem.

See below PR for reference:
https://github.com/kubernetes-sigs/descheduler/pull/523
 2021-04-01 00:28:20 -05:00