Skip to content

Latest commit

 

History

History
129 lines (88 loc) · 5.16 KB

lab7.md

File metadata and controls

129 lines (88 loc) · 5.16 KB
Raw

Lab 7 - Asynchronous functions

Before starting this lab, create a new folder for your files:

$ mkdir -p lab7 \
   && cd lab7

Call a function synchronously vs asynchronously

When you call a function synchronously a connection is made to the gateway through to your function and is held open for the whole execution time. Synchronous calls are blocking so you should see your client pause and become inactive until the function has completed its task.

  • The gateway uses a route of: /function/<function_name>
  • You have to wait until it has finished
  • You get the result after the call
  • You know if it passed or failed

Asynchronous tasks run in a similar way with a few differences:

  • The gateway uses a different route: /async-function/<function_name>
  • The client gets an immediate response of 202 Accepted from the gateway
  • The function is invoked later using a queue-worker
  • By default the result is discarded

Let's try a quick demo.

$ faas-cli new --lang dockerfile long-task --prefix="<your-docker-username-here>"

Edit long-task/Dockerfile and change the fprocess to sleep 1.

Now build, deploy and invoke your function 10 times synchronously like this:

echo -n "" | faas-cli invoke long-task
echo -n "" | faas-cli invoke long-task
echo -n "" | faas-cli invoke long-task
echo -n "" | faas-cli invoke long-task
echo -n "" | faas-cli invoke long-task
echo -n "" | faas-cli invoke long-task
echo -n "" | faas-cli invoke long-task
echo -n "" | faas-cli invoke long-task
echo -n "" | faas-cli invoke long-task
echo -n "" | faas-cli invoke long-task

Now invoke the function 10 times asynchronously:

echo -n "" | faas-cli invoke long-task --async
echo -n "" | faas-cli invoke long-task --async
echo -n "" | faas-cli invoke long-task --async
echo -n "" | faas-cli invoke long-task --async
echo -n "" | faas-cli invoke long-task --async
echo -n "" | faas-cli invoke long-task --async
echo -n "" | faas-cli invoke long-task --async
echo -n "" | faas-cli invoke long-task --async
echo -n "" | faas-cli invoke long-task --async
echo -n "" | faas-cli invoke long-task --async
echo -n "" | faas-cli invoke long-task --async

What did you observe? The first example should have taken 10 seconds where as the second example would have returned to your prompt within around a second or less. The work will still take 10x1 seconds to complete, but that is now going to be placed on a queue for deferred execution.

Asynchronous function calls are well-suited to tasks where you can defer the execution until a later time, or you don't need the result on the client.

A good example of this may be when receiving webhooks from GitHub - there may be a maximum processing time which GitHub will allow your connection to remain open for, an asynchronous call accepts the work and returns immediately.

View the queue-worker's logs

The default stack for OpenFaaS uses NATS Streaming for queueing and deferred execution. You can view the logs with the following command:

Docker Swarm

docker service logs -f func_queue-worker

Kubernetes

kubectl logs deployment/queue-worker -n openfaas

Use an X-Callback-Url with requestbin

If you need a result from a function called asynchronously you have two options:

  • Alter its code to notify an endpoint or messaging system with its result

This option may not be suitable in all circumstances and involves writing additional code.

  • Make use of the built-in behaviour of callbacks

The built-in callbacks allow a call to a function to provide a URL where the queue-worker will automatically report the success or failure of a function along with the result. Some additional request headers are sent to the callback, for a complete list see Callback request headers

Head over to requestbin and create a new "bin" - this will be a URL on the public internet that can receive your function's result.

For the purpose of this lab, be sure to uncheck the 'Private' checkbox as this will save you from needing to log in.

https://requestbin.com/

Now copy the "Bin URL" and paste it below:

For example (http://requestbin.com/r/1i7i1we1)

$ echo -n "LaterIsBetter" | faas-cli invoke figlet --async --header "X-Callback-Url=http://requestbin.com/r/1i7i1we1"

Now refresh the page on the requestbin site and you will see the result from figlet:

 _          _           ___     ____       _   _            
| |    __ _| |_ ___ _ _|_ _|___| __ )  ___| |_| |_ ___ _ __ 
| |   / _` | __/ _ \ '__| |/ __|  _ \ / _ \ __| __/ _ \ '__|
| |__| (_| | ||  __/ |  | |\__ \ |_) |  __/ |_| ||  __/ |   
|_____\__,_|\__\___|_| |___|___/____/ \___|\__|\__\___|_|

Pro-tip: It is also possible to use another function as the X-Callback-Url - this would be great for notifying yourself over Slack or Email when an asynchronous workload has been processed. To call another function with the result set the X-Callback-Url to http://gateway:8080/function/<function_name>.

Now move onto Lab 8