d1a89cf5ed
Put risky NodeToRemove in the end of needDrain list
2022-12-06 15:11:16 +00:00
94f1920265
Use PdbRemainingDisruptions in Planner
2022-12-06 15:11:08 +00:00
10d3f25996
Use scheduling package in filterOutSchedulable processor
2022-11-23 12:32:59 +00:00
92bba5c93f
Allow forking snapshot more than 1 time
2022-11-16 14:19:30 +00:00
a703c3f6cf
Merge pull request #5230 from x13n/scaledown6
...
Base parallel scale down implementation
2022-10-25 03:18:35 -07:00
accf58f36c
Base parallel scale down implementation
2022-10-24 20:14:48 +02:00
585ad02c04
Remove dead code for handling simulation errors
2022-10-20 14:20:35 +02:00
92f5b8673e
Extract scheduling hints to a dedicated object
...
This removes the need for passing maps back and forth when doing
scheduling simulations.
2022-10-20 11:44:15 +02:00
18f2e67c4f
Split out code from simulator package
2022-10-18 11:51:44 +02:00
0ee2a359e7
Add option to wait for a period of time after node tainting/cordoning
...
Node state is refreshed and checked again before deleting the node
It gives kube-scheduler time to acknowledge that nodes state has
changed and to stop scheduling pods on them
2022-10-13 10:37:56 +03:00
28548705f8
Allow simulator to persist changes in cluster snapshot
2022-09-15 16:18:57 +02:00
c127763a45
switched policy for PodDisruptionBudget from v1beta1 to v1 in time for 1.25
2022-06-24 19:13:03 +02:00
07a8d38933
Allow checking specific nodes removal in simulator
...
Wraps simulator functions into an object, so that there's no need to
pass around parametrs that are not changing between CA loop iterations.
2022-05-06 15:49:34 +02:00
358f3a9218
Extract utilization info to a separate package
2022-04-26 08:48:45 +02:00
8064d6d1fd
Introduce the scale down processor that picks the final scale down candidates.
2022-01-03 16:05:36 +00:00
2529b667c5
ignoring terminated pods in scaledown
2020-12-21 16:22:50 +01:00
9835d59e28
Merge pull request #3749 from yaroslava-serdiuk/DS_utilization
...
subtract mirror pod utilization with ds utilization from node allocatable
2020-12-16 05:36:22 -08:00
8a94cec938
substract mirror pod utilization with daemonset utilization from node allocatable + modify tests
2020-12-16 10:49:41 +00:00
aa9367f56f
Merge pull request #3701 from yaroslava-serdiuk/DS_evict
...
add daemonset eviction for non-empty nodes
2020-11-30 05:02:49 -08:00
dcf0970208
add daemonset eviction for non-empty nodes
2020-11-30 11:53:16 +00:00
0fb897b839
Update imports after scheduler scheduler/framework/v1alpha1 removal
2020-11-30 10:48:52 +00:00
d945ca7eed
Merge pull request #3641 from yaroslava-serdiuk/DS2
...
change how DaemonSet utilization is calculated
2020-10-29 10:52:06 -07:00
8480c79ce0
change how DaemonSet utilization is calculated
2020-10-23 14:39:24 +00:00
1f95298899
Ensure CA simulator takes existing nodes into account in findPlaceFor method
2020-10-16 15:14:26 +02:00
2ae13f561b
Use FitsAny in drain simulation
...
This is the scale-down equivalent of #3429 and it speeds-up
findUnneeded by 5x+ in very large clusters (by reducing the
number of expensive PreFilter calls #nodes times).
A side effect of this change is removing
"Simulating scheduling of <pod> to <node> return error <error>"
logs. Using FitsAny we no longer have per-node scheduler errors
that we could log. I think that's actually a good thing - even
with klogx this log was incredibly spammy in cluster with >100
nodes and its practical value was questionable.
2020-09-02 14:26:02 +02:00
9fb6cdc079
Fix go fmt errors
2020-06-08 13:52:24 +02:00
2160e6d49e
Rewrite glogx to work with klogv2 (+rename klogx)
2020-06-05 17:22:26 +02:00
655b4081f4
Migrate to klog v2
2020-06-05 17:22:26 +02:00
73a5cdf928
Address recent breaking changes in scheduler
...
The following things changed in scheduler and needed to be fixed:
* NodeInfo was moved to schedulerframework
* Some fields on NodeInfo are now exposed directly instead of via getters
* NodeInfo.Pods is now a list of *schedulerframework.PodInfo, not *apiv1.Pod
* SharedLister and NodeInfoLister were moved to schedulerframework
* PodLister was removed
2020-04-24 17:54:47 +02:00
14c5ca44a8
fix typo in snapshot var name
2020-03-10 11:36:57 +01:00
9c5507eb95
pass list of destination names, not map
2020-03-09 14:15:16 +01:00
4691f1ccea
skip node if node info not found
2020-03-09 13:45:11 +01:00
38a920c423
pass names of destinations to find nodes to remove
2020-03-05 16:09:51 +01:00
d53cd35818
pass node names instead of node infos to find nodes to remove
2020-03-05 12:07:36 +01:00
0b7c45e88a
stop passing scheduled pods around
2020-03-03 16:23:49 +01:00
d11b39603d
pass node infos to simulator
2020-03-03 16:23:49 +01:00
5d44b202bc
Forget FakeNodeInfoForNodeName ever existed
2020-02-21 15:36:21 +01:00
af1dd84305
Merge pull request #2799 from aleksandra-malinowska/delta-snapshot-4
...
Add delta snapshot implementation
2020-02-14 09:20:17 -08:00
7a188ab50d
Provide ScaleDownStatusProcessor with info about unremovable nodes
2020-02-11 15:27:33 +01:00
de631034c0
Pass node name when removing pod from snapshot
2020-02-10 17:28:06 +01:00
7b67d3f582
klog.Fatalf on error from ClusterSnapshot.Revert()
2020-02-04 20:52:07 +01:00
d7770e3044
Use ClusterSnapshot in ScaleDown
2020-02-04 20:51:48 +01:00
6b2287af4f
Pass ClusterSnaphost explicitly to PredicateChecker
2020-02-04 20:51:24 +01:00
373c558303
Extract PredicateChecker interface
2020-02-04 20:51:18 +01:00
b0c6d25182
Cleanup simulator.PredicateError
2020-02-04 20:51:11 +01:00
4a2b8c7dfc
Remove use of PredicateMetadata
2020-02-04 20:51:05 +01:00
b0f05c6542
Uniformize node shuffling when finding nodes to remove
...
Previously, the shuffling was biased
2020-01-31 23:28:32 -08:00
3470a9f1c0
Migrate scheduler creation to scheduler.New()
2019-11-22 14:07:29 +01:00
f0fbf7a87a
[cluster-autoscaler] Allow “custom” DaemonSet pods
...
* Create `utils/pod` package: pod kind detection functions.
* Update DaemonSet Pod detection logic: introduce the annotation
`cluster-autoscaler.kubernetes.io/daemonset-pod` to declare a Pod as a
DaemonSet Pod.
* Update `simulator` package to use the new `utils/pod` package function.
* Cleanup `getRequiredPodsForNode()` function.
Signed-off-by: cedric lamoriniere <cedric.lamoriniere@datadoghq.com>
2019-11-06 11:23:23 +01:00
dc64d0aab2
Adding ScaleDownNodeProcessor
2019-08-12 20:19:55 +02:00
Yaroslava Serdiuk
Bartłomiej Wróblewski
Kubernetes Prow Robot
Daniel Kłobuszewski
Alexandru Matei
mikelo
Daniel Gutowski
david.benque
Vivek Bagade
Maciek Pytel
Jakub Tużnik
Aleksandra Malinowska
Łukasz Osipiuk
Taylor Barrella
cedric lamoriniere