This message was deleted.

no, we haven’t change any of the manifests sync behavior in ages

Thank you @creamy-pencil-82913 for reply Not sure what am i missing in 1.26, using same config in in 1.23 and 1.26

In K3s 1.23- it is working as expected
 
ps-k8s-test helm chart yaml file is in /var/lib/rancher/k3s/server/manifests 

[root@test1 ~]# kubectl get pods -A | grep test
kube-system           helm-install-ps-k8s-test-misc-tctpv                  0/1     Completed   0             15m

Now Deleting the ps-k8s-test job
[root@test1 ~]# kubectl delete job helm-install-ps-k8s-test-misc -n kube-system
job.batch "helm-install-ps-k8s-test-misc" deleted

then ps-k8s-test helm install job created immediately since respective helm chart in manifest folder 
[root@test1 ~]# kubectl get pods -A | grep test
kube-system           helm-install-ps-k8s-test-misc-2mhrz                  0/1     ContainerCreating   0             2s

********************************

But k3s 1.26.7 its not creating deleted job k8s resources autimatically

[root@trest2 ~]# kubectl get pods -A | grep test
kube-system           helm-delete-ps-k8s-test-misc-qcfdb                   0/1     Completed   0          51m
[root@trest2 ~]#

Deleting job 

[root@trest2 ~]# kubectl delete job helm-delete-ps-k8s-test-misc -n kube-system
job.batch "helm-delete-ps-k8s-test-misc" deleted
[root@trest2 ~]#
[root@trest2 ~]# kubectl get pods -A | grep test
[root@trest2 ~]#


[root@trest2 ~]# kubectl get pods -A | grep test
[root@trest2 ~]#

[root@test2  ~]# kubectl get pods -A | grep test

it did not recreated the job automatically. it is creating it when we restart k3s though.

There was a bug in a couple quite old releases where it would continually reapply the manifests. it was not intended to do that, it was intended to just apply them when they change, by tracking the file checksum and mtime. If you’re on 1.23, which is quite old, its possible you’re on one of the releases that had the behavior of constantly applying - which puts more load on the server.

To make sure I understand it correct, 1.26 is working as expected. it only creates the resources when file checksum amd modified time changes? is