mirror of
https://github.com/slurm-personal/school-dev-k8s.git
synced 2026-06-27 13:50:24 +00:00
Slurm
198 lines
3.3 KiB
Markdown
198 lines
3.3 KiB
Markdown
# Job
|
|
|
|
### Запускаем простой job
|
|
|
|
1) Создаем job
|
|
|
|
```bash
|
|
kubectl apply -f ~/school-dev-k8s/practice/10.oneshot-tasks/1.job/job.yaml
|
|
```
|
|
|
|
2) Проверяем
|
|
|
|
```bash
|
|
kubectl get job
|
|
```
|
|
|
|
Видим:
|
|
|
|
```bash
|
|
NAME COMPLETIONS DURATION AGE
|
|
hello 1/1 2s 88s
|
|
```
|
|
|
|
3) Смотрим на Pod'ы
|
|
|
|
```bash
|
|
kubectl get pod
|
|
```
|
|
|
|
Видим Pod, созданный Job'ой:
|
|
|
|
```bash
|
|
NAME READY STATUS RESTARTS AGE
|
|
hello-6l9tv 0/1 Completed 0 2m6s
|
|
```
|
|
|
|
4) Смотрим его логи
|
|
|
|
```bash
|
|
kubectl logs hello-6l9tv
|
|
```
|
|
|
|
Видим что все отработало правильно:
|
|
|
|
```bash
|
|
Mon Mar 18 15:06:10 UTC 2019
|
|
Hello from the Kubernetes cluster
|
|
```
|
|
|
|
5) Удаляем Job
|
|
|
|
```bash
|
|
kubectl delete job hello
|
|
```
|
|
|
|
### Проверяем работу параметра backoffLimit
|
|
|
|
6) Открываем файл job.yaml и находим командy выполняющуюся в Pod:
|
|
|
|
```yaml
|
|
args:
|
|
- /bin/sh
|
|
- -c
|
|
- date; echo Hello from the Kubernetes cluster
|
|
```
|
|
|
|
И ломаем полностью:
|
|
|
|
```yaml
|
|
args:
|
|
- /bin/sh
|
|
- -c
|
|
- date; echo Hello from the Kubernetes cluster; exit 1
|
|
```
|
|
|
|
7) Создаем Job:
|
|
|
|
```bash
|
|
kubectl apply -f ~/school-dev-k8s/practice/10.oneshot-tasks/1.job/job.yaml
|
|
```
|
|
|
|
8) Проверяем
|
|
|
|
```bash
|
|
kubectl get job
|
|
```
|
|
|
|
Видим:
|
|
|
|
```bash
|
|
NAME COMPLETIONS DURATION AGE
|
|
hello 0/1 27s 27s
|
|
```
|
|
|
|
9) Смотрим на Pod'ы
|
|
|
|
```bash
|
|
kubectl get pod
|
|
```
|
|
|
|
Видим Pod'ы, созданные Job'ой:
|
|
|
|
```bash
|
|
NAME READY STATUS RESTARTS AGE
|
|
hello-5nvqf 0/1 Error 0 22m
|
|
hello-ks4ks 0/1 Error 0 22m
|
|
hello-rl984 0/1 Error 0 22m
|
|
```
|
|
|
|
Они в статусе Error
|
|
|
|
10) Смотрим в описание Job'ы
|
|
|
|
```bash
|
|
kubectl describe job hello
|
|
```
|
|
|
|
Видим, что backoffLimit сработал
|
|
|
|
```bash
|
|
Warning BackoffLimitExceeded 23m job-controller Job has reached the specified backoff limit
|
|
```
|
|
|
|
11) Удаляем Job
|
|
|
|
```bash
|
|
kubectl delete job hello
|
|
```
|
|
|
|
### Проверяем работу параметра activeDeadlineSeconds
|
|
|
|
12) Открываем файл job.yaml и находим командy, выполняющуюся в Pod'е:
|
|
|
|
```yaml
|
|
args:
|
|
- /bin/sh
|
|
- -c
|
|
- date; echo Hello from the Kubernetes cluster
|
|
```
|
|
|
|
И делаем ее бесконечной
|
|
|
|
```yaml
|
|
args:
|
|
- /bin/sh
|
|
- -c
|
|
- while true; do date; echo Hello from the Kubernetes cluster; sleep 1; done
|
|
```
|
|
|
|
13) Создаем Job
|
|
|
|
```bash
|
|
kubectl apply -f ~/school-dev-k8s/practice/10.oneshot-tasks/1.job/job.yaml
|
|
```
|
|
|
|
14) Проверяем
|
|
|
|
```bash
|
|
kubectl get job
|
|
```
|
|
|
|
Видим:
|
|
|
|
```bash
|
|
NAME COMPLETIONS DURATION AGE
|
|
hello 0/1 27s 27s
|
|
```
|
|
|
|
15) Смотрим на Pod'ы
|
|
|
|
```bash
|
|
kubectl get pod
|
|
```
|
|
|
|
Видим Pod'ы, созданный Job'ой
|
|
|
|
```bash
|
|
NAME READY STATUS RESTARTS AGE
|
|
hello-bt6g6 1/1 Running 0 5s
|
|
```
|
|
|
|
16) Ждем 60 секунд и проверяем Job
|
|
|
|
```bash
|
|
kubectl describe job hello
|
|
```
|
|
|
|
Видим, что activeDeadlineSeconds сработал
|
|
```bash
|
|
Warning DeadlineExceeded 2m17s job-controller Job was active longer than specified deadline
|
|
```
|
|
|
|
17) Удаляем Job
|
|
|
|
```bash
|
|
kubectl delete job hello
|
|
```
|