Parallel Processing using Expansions
Example: Multiple Job Objects from Template Expansion
In this example, we will run multiple Kubernetes Jobs created from a common template. You may want to be familiar with the basic, non-parallel, use of Jobs first.
Basic Template Expansion
First, download the following template of a job to a file called job.yaml
job.yaml
|
|---|
|
Unlike a pod template, our job template is not a Kubernetes API type. It is just
a yaml representation of a Job object that has some placeholders that need to be filled
in before it can be used. The $ITEM syntax is not meaningful to Kubernetes.
In this example, the only processing the container does is to echo a string and sleep for a bit.
In a real use case, the processing would be some substantial computation, such as rendering a frame
of a movie, or processing a range of rows in a database. The $ITEM parameter would specify for
example, the frame number or the row range.
This Job and its Pod template have a label: jobgroup=jobexample. There is nothing special
to the system about this label. This label
makes it convenient to operate on all the jobs in this group at once.
We also put the same label on the pod template so that we can check on all Pods of these Jobs
with a single command.
After the job is created, the system will add more labels that distinguish one Job’s pods
from another Job’s pods.
Note that the label key jobgroup is not special to Kubernetes. You can pick your own label scheme.
Next, expand the template into multiple files, one for each item to be processed.
# Expand files into a temporary directory
$ mkdir ./jobs
$ for i in apple banana cherry
do
cat job.yaml | sed "s/\$ITEM/$i/" > ./jobs/job-$i.yaml
done
Check if it worked:
$ ls jobs/
job-apple.yaml
job-banana.yaml
job-cherry.yaml
Here, we used sed to replace the string $ITEM with the loop variable.
You could use any type of template language (jinja2, erb) or write a program
to generate the Job objects.
Next, create all the jobs with one kubectl command:
$ kubectl create -f ./jobs
job "process-item-apple" created
job "process-item-banana" created
job "process-item-cherry" created
Now, check on the jobs:
$ kubectl get jobs -l jobgroup=jobexample
NAME DESIRED SUCCESSFUL AGE
process-item-apple 1 1 31s
process-item-banana 1 1 31s
process-item-cherry 1 1 31s
Here we use the -l option to select all jobs that are part of this
group of jobs. (There might be other unrelated jobs in the system that we
do not care to see.)
We can check on the pods as well using the same label selector:
$ kubectl get pods -l jobgroup=jobexample --show-all
NAME READY STATUS RESTARTS AGE
process-item-apple-kixwv 0/1 Completed 0 4m
process-item-banana-wrsf7 0/1 Completed 0 4m
process-item-cherry-dnfu9 0/1 Completed 0 4m
There is not a single command to check on the output of all jobs at once, but looping over all the pods is pretty easy:
$ for p in $(kubectl get pods -l jobgroup=jobexample --show-all -o name)
do
kubectl logs $p
done
Processing item apple
Processing item banana
Processing item cherry
Multiple Template Parameters
In the first example, each instance of the template had one parameter, and that parameter was also used as a label. However label keys are limited in what characters they can contain.
This slightly more complex example uses the jinja2 template language to generate our objects. We will use a one-line python script to convert the template to a file.
First, copy and paste the following template of a Job object, into a file called job.yaml.jinja2:
{%- set params = [{ "name": "apple", "url": "http://www.orangepippin.com/apples", },
{ "name": "banana", "url": "https://en.wikipedia.org/wiki/Banana", },
{ "name": "raspberry", "url": "https://www.raspberrypi.org/" }]
%}
{%- for p in params %}
{%- set name = p["name"] %}
{%- set url = p["url"] %}
apiVersion: batch/v1
kind: Job
metadata:
name: jobexample-{{ name }}
labels:
jobgroup: jobexample
spec:
template:
metadata:
name: jobexample
labels:
jobgroup: jobexample
spec:
containers:
- name: c
image: busybox
command: ["sh", "-c", "echo Processing URL {{ url }} && sleep 5"]
restartPolicy: Never
---
{%- endfor %}
The above template defines parameters for each job object using a list of
python dicts (lines 1-4). Then a for loop emits one job yaml object
for each set of parameters (remaining lines).
We take advantage of the fact that multiple yaml documents can be concatenated
with the --- separator (second to last line).
.) We can pipe the output directly to kubectl to
create the objects.
You will need the jinja2 package if you do not already have it: pip install --user jinja2.
Now, use this one-line python program to expand the template:
alias render_template='python -c "from jinja2 import Template; import sys; print(Template(sys.stdin.read()).render());"'
The output can be saved to a file, like this:
cat job.yaml.jinja2 | render_template > jobs.yaml
Or sent directly to kubectl, like this:
cat job.yaml.jinja2 | render_template | kubectl create -f -
Alternatives
If you have a large number of job objects, you may find that:
- Even using labels, managing so many Job objects is cumbersome.
- You exceed resource quota when creating all the Jobs at once, and do not want to wait to create them incrementally.
- You need a way to easily scale the number of pods running concurrently. One reason would be to avoid using too many compute resources. Another would be to limit the number of concurrent requests to a shared resource, such as a database, used by all the pods in the job.
- Very large numbers of jobs created at once overload the Kubernetes apiserver, controller, or scheduler.
In this case, you can consider one of the other job patterns.