This is an automated email from the ASF dual-hosted git repository.
mani pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/yunikorn-site.git
The following commit(s) were added to refs/heads/master by this push:
new 5b0ddd23de [YUNIKORN-3244] Document improved victim selection process
(#548)
5b0ddd23de is described below
commit 5b0ddd23de431a1dac116bc366eda1a5c02a4dfa
Author: Manikandan R <[email protected]>
AuthorDate: Mon Apr 6 09:01:13 2026 +0530
[YUNIKORN-3244] Document improved victim selection process (#548)
Closes: #548
Signed-off-by: Manikandan R <[email protected]>
---
docs/design/quota_preemptor.md | 17 +++++++++++------
1 file changed, 11 insertions(+), 6 deletions(-)
diff --git a/docs/design/quota_preemptor.md b/docs/design/quota_preemptor.md
index 2edc8e6fe4..8adcba6291 100644
--- a/docs/design/quota_preemptor.md
+++ b/docs/design/quota_preemptor.md
@@ -151,15 +151,20 @@ Whenever any leaf queue quota changes, the above
described preemptable resource
Collect all running tasks based on below described criteria:
1. Ignore Daemon set pods
-2. Choose pods which have “at least one” match with preemptable resources.
+2. Choose pods only whose resources have “at least one type” match with
preemptable resources. Otherwise, completely "dis-joint" pods resource type
wise are ignored.
Once the set of tasks been collected, same has to be sorted based on the below
described preference rules:
-1. Prefer Non originator pods over originator
-2. In case of tie in rule \#1, prefer pods based on priority. Lower priority
picked up first followed by high priority pods.
-3. In case of tie in rule \#2, Prefer pods for which the “allowPreemption”
flag has been set to true over other pods.
-4. In case of tie in rule \#3, prefer “young” pods based on the age
-
+1. Prefer the highest scoring pod over the others. Scores are computed based
on the following criteria:
+ 1. Prefer Non originator pods over originator. Score is higher for Non
originator pods when compared to originator.
+ 2. Prefer pods for which the “allowPreemption” flag has been set to true
over other pods. Score is higher for pods for which “allowPreemption” flag has
been set when compared to others.
+ 3. Prefer pods based on "type matching" percentage with preemptable
resources. Higher the "type matching" percentage, higher the score. "type
matching" percentage is calculated based on the no. of matching resource types
against preemptable resource types and total no. of resource types used by the
pod. For example, pod used only one resource type and that too matching with
preemptable resource types carries 100% type matching percentage.
+
+Once the total score derived for each pod, prefer the highest scoring pod over
the others.
+
+2. In case of tie or having same scores in rule \#1, prefer pods based on
priority. Lower priority picked up first followed by high priority pods.
+3. In case of tie in rule \#2, prefer “young” pods based on the age. Pod
creation timestamp is truncated upto Unit "Hour" and used as lowest
granularity. So, pods created in same hour are not really ordered, whereas pods
created in different hours are ordered effectively by not taking seconds,
micro, millis etc. into account.
+4. In case of tie in rule \#3, prefer the least used pods resource wise. Order
pods based on allocated resources. Only the preemptable resource types used in
pods are taken into account.
Once the tasks are sorted, tasks can be traversed one by one and preempted to
free up resources until we reach the target.
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
