You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: README.md
+15-1Lines changed: 15 additions & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -15,6 +15,20 @@ Run the descheduler in your OpenShift cluster to move pods based on specific str
15
15
| 5.1.3 | 4.17, 4.18 | 1.31 | 1.22 |
16
16
| 5.2.0 | 4.19, 4.20 | 1.32 | 1.23 |
17
17
18
+
## Rebase instruction
19
+
20
+
```
21
+
Steps:
22
+
- [ ] bump .ci-operator.yaml if needed (as a separate PR and wait until the changes gets propagated to https://github.com/openshift/release/tree/master/ci-operator/config/openshift/ of the corresponding CI definition)
23
+
- [ ] bump go version in a go.mod file if needed (check go.mod of the corresponding kubernetes release under https://github.com/kubernetes/kubernetes/branches)
24
+
- [ ] bump all k8s.io/*, github.com/openshift/* and other relevant dependencies (you can consults the corresponding go.mod file as mentioned previously)
25
+
- [ ] (recommended) commit all the go.mod and go.sum changes separatelly from `go mod vendor` changes
26
+
- [ ] run "go mod vendor" and commit the changes
27
+
- [ ] build the code (e.g. by running make) and adjust the code if needed to make the building step compile successfully
28
+
- [ ] run unit tests (e.g. by running make unit-tests) successfully
29
+
- [ ] commit the code adjustments if there are any
30
+
```
31
+
18
32
## Deploy the operator
19
33
20
34
### Quick Development
@@ -193,7 +207,7 @@ The profile exposes the following customization:
193
207
By default, this profile will enable load-aware descheduling based on the `PrometheusCPUCombined` Prometheus query.
194
208
By default, the thresholds will be dynamic (based on the distance from the average utilization) and asymmetric (all the nodes below the average will be considered as underutilized to help rebalancing overutilized outliers) tolerating low deviations (10%).
195
209
196
-
By default, this profile configures the descheduler to restrict the maximum number of overall parallel evictions to 5 and
210
+
By default, this profile configures the descheduler to restrict the maximum number of overall parallel evictions to 5 and
197
211
the maximum number of evictions per node to 2 aligning with KubeVirt defaults around concurrent live migrations.
198
212
Those two values can be customized with `evictionLimits.total` and `evictionLimits.node` parameters.
0 commit comments