ps:restart should rolling restart the pods

[dmcnaught]

I read this - and it says resolved:

deis/deis#3148

then I tested w deis 2.21.6 on K8s 1.17 (installed in AWS w kops) and it doesn't do a rolling restart:

NAME                        READY   STATUS              RESTARTS   AGE
test-cmd-85c69bf65f-4fpct   0/1     Running             0          6s
test-cmd-85c69bf65f-dcdqq   0/1     Running             0          6s
test-cmd-85c69bf65f-ftgr4   0/1     Terminating         0          51s
test-cmd-85c69bf65f-q4c5s   0/1     ContainerCreating   0          6s
test-cmd-85c69bf65f-sthkt   0/1     Running             0          6s
test-cmd-85c69bf65f-vj5cm   0/1     Terminating         0          51s

[Cryptophobia]

Is this still happening? It could be logic in workflow-cli that we need to change or the logic inside the controller when making the deployment object.

[dmcnaught]

Yes, it is still happening on K8s v1.18.12 with deis v2.22.1 (plus some fixes that make it like 2.22.3)

» kubectl -n test get po
NAME                       READY   STATUS    RESTARTS   AGE
test-cmd-554954c65-28pbd   1/1     Running   0          2m1s
test-cmd-554954c65-9fjnm   1/1     Running   0          2m
test-cmd-554954c65-bxsvl   1/1     Running   0          2m
test-cmd-554954c65-kw4lw   1/1     Running   0          2m
test-cmd-554954c65-mbv2j   1/1     Running   0          2m
» deis ps:restart -a test
...
» kubectl -n test get po
NAME                       READY   STATUS        RESTARTS   AGE
test-cmd-554954c65-28pbd   0/1     Terminating   0          2m11s
test-cmd-554954c65-8njdq   0/1     Running       0          3s
test-cmd-554954c65-bq8nr   0/1     Running       0          3s
test-cmd-554954c65-jzpm7   0/1     Running       0          3s
test-cmd-554954c65-k82zr   0/1     Running       0          3s
test-cmd-554954c65-kw4lw   0/1     Terminating   0          2m10s
test-cmd-554954c65-v8gmz   0/1     Running       0          3s

No healthchecks are passing - it restarted all the containers.