Commit Graph

31 Commits

Author SHA1 Message Date
Davanum Srinivas
7b8c9acc09
remove unused code
Change-Id: If821920ec8872e326b7d85437ad8d2620807799d
2019-04-19 08:36:31 -04:00
Wei Huang
7afbd68730
add metrics to record number of pending pods in different queues 2019-04-08 17:51:07 -07:00
Ted Yu
4b4c20be7c Use read lock for PendingPods 2019-04-06 13:48:51 -07:00
Kubernetes Prow Robot
52ec2a0009
Merge pull request #75497 from goodluckbot/remove-backoff-util
Integrate backoff mechanism into the scheduling queue and remove the …
2019-04-05 18:06:41 -07:00
goodluckbot
151649df4c Integrate backoff mechanism into the scheduling queue and remove the Backoff util 2019-04-06 00:15:00 +08:00
wangqingcan
8fd45d8300 clean up func podTimestamp 2019-03-27 11:05:19 +08:00
Wei Huang
2b554b4286
scheduler: cleanup unused parameters in unschedulableQ 2019-03-13 11:54:16 -07:00
Jonathan Basseri
df4d65d2e1 Ensure that pods obey backoff timers.
The function AddUnschedulableIfNotPresent is responsible for
initializing or updating backoff timers for pods that could not be
scheduled. The helper function backoffPod does that work, but was not
being called in all cases.

This moves that call to be (mostly) unconditional, while cleaning up
comments and error handling.
2019-02-26 10:21:18 -08:00
wangqingcan
ea9e1a4118 not updae timestamp for each scheduling attempt 2019-02-22 09:46:18 +08:00
Yecheng Fu
ba47beffd2 Fix weakness of current receivedMoveRequest
- add incremental scheduling cycle
- instead of set a flag on move reqeust, we cache current scheduling
cycle in moveRequestCycle
- when unschedulable pods are added back, compare its cycle with
moveRequestCycle to decide whether it should be added into active queue
or not
2019-01-30 10:13:49 +08:00
Wei Huang
02cdc24cd3
enforce unscheduable pod has proper chance to be retried 2019-01-22 09:23:51 +08:00
Kubernetes Prow Robot
d857790d36
Merge pull request #72558 from denkensk/add-goroutine-move-unschedulablepods-to-activeq
Move unschedulable pods to the active queue if they are not retried for more than 1 minute
2019-01-16 17:15:16 -08:00
wangqingcan
de8cfdcd79 add goroutine to move unschedulablepods to activeq regularly 2019-01-16 12:08:19 +08:00
Kubernetes Prow Robot
db60e4f20c
Merge pull request #72632 from danielqsj/sch_pod
Move getNextPod to closure
2019-01-08 19:22:19 -08:00
Kubernetes Prow Robot
5a708017e9
Merge pull request #72619 from everpeace/fix-starvation-by-sorting-queue-with-last-probe-time
change sort function of scheduling queue to avoid starvation when a lot of unscheduleable pods are in the queue
2019-01-07 20:24:24 -08:00
danielqsj
8dc1e83946 Add func comment for MakeNextPodFunc 2019-01-08 11:07:47 +08:00
danielqsj
554ad383e3 Move getNextPod to closure 2019-01-07 17:54:49 +08:00
Shingo Omura
22079a79d4 change sort function of scheduling queue to avoid starvation when unschedulable pods are in the queue
When starvation heppens:
- a lot of unschedulable pods exists in the head of queue
- because condition.LastTransitionTime is updated only when condition.Status changed
- (this means that once a pod is marked unschedulable, the field never updated until the pod successfuly scheduled.)

What was changed:
- condition.LastProbeTime is updated everytime when pod is determined
unschedulable.
- changed sort function so to use LastProbeTime to avoid starvation
described above

Consideration:
- This changes increases k8s API server load because it updates Pod.status whenever scheduler decides it as
unschedulable.

Signed-off-by: Shingo Omura <everpeace@gmail.com>
2019-01-07 11:05:45 +09:00
Kubernetes Prow Robot
222d7837a8
Merge pull request #72303 from cofyc/fix72013-cleanup-scheduler
Update comments to reflect current logic
2019-01-03 15:51:34 -08:00
Yecheng Fu
fa88bcc154 Update comments for PriorityQueue.
Log condition reason which is useful for debugging.
2019-01-03 09:25:05 +08:00
Bobby (Babak) Salamat
7044145920 Fix race in setting nominated node 2018-12-29 22:54:15 -08:00
Bobby (Babak) Salamat
48b6f75829 Add pods in the backoff queue to the list of pending pods 2018-12-20 17:28:23 -08:00
Jonathan Basseri
d27d28a44e Flatten nominated pod logic in PriorityQueue.
This replaces deeply nested ifs & fors with early returns & continues.
2018-12-14 18:11:12 -08:00
Jonathan Basseri
fae4f69d36 Fix return value of PriorityQueue.Add.
This function was returning a non-nil error for the common, non-failure
case. The fix is to properly scope local error values and add early
returns.
2018-12-14 16:44:21 -08:00
mlmhl
2fe9b1438c activate unschedulable pods only if the node became more schedulable 2018-12-10 09:48:23 +08:00
Gregory Haynes
5e4ccede4c Reschedule with backoff
With the alpha scheduling queue we move pods from unschedulable to
active on certain events without a backoff. As a result we can cause
starvation issues if high priority pods are in the unschedulable queue.
Implement a backoff mechanism for pods being moved to active.

Closes #56721
2018-12-06 16:38:34 +00:00
Gregory Haynes
c821f2ed2f Move scheduling Heap in to scheduler.core.utils
The Heap data structure is useful for our backoff system in addition to
scheduling queue. Move it to somewhere it can be consumed by both
systems and properly export needed names. Also adding unit tests
from client-go/tools/cache/heap.go.
2018-12-04 18:01:42 +00:00
Bobby (Babak) Salamat
554acf2b38 Change sort function of the scheduling queue to avoid starvation 2018-11-27 17:08:40 -08:00
Wei Huang
b4fd11512a
ensure scheduler preemptor behaves in an efficient/correct path
- don't update nominatedMap cache when Pop() an element from activeQ
- instead, delete the nominated info from cache when it's "assumed"
- unit test behavior adjusted
- expose SchedulingQueue in factory.Config
2018-11-16 14:22:15 -08:00
Davanum Srinivas
954996e231
Move from glog to klog
- Move from the old github.com/golang/glog to k8s.io/klog
- klog as explicit InitFlags() so we add them as necessary
- we update the other repositories that we vendor that made a similar
change from glog to klog
  * github.com/kubernetes/repo-infra
  * k8s.io/gengo/
  * k8s.io/kube-openapi/
  * github.com/google/cadvisor
- Entirely remove all references to glog
- Fix some tests by explicit InitFlags in their init() methods

Change-Id: I92db545ff36fcec83afe98f550c9e630098b3135
2018-11-10 07:50:31 -05:00
Wei Huang
9da576f03c
move SchedulingQueue to pkg/scheduler/internal/queue 2018-09-28 11:51:02 -07:00