Skip to content

Pc devoxx #8

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 2 commits into
base: start
Choose a base branch
from
Open

Pc devoxx #8

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
bff5965
Clean tp3
pedroelmoucho Oct 4, 2023
0fdf200
review doc
pedroelmoucho Oct 4, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 5 additions & 2 deletions doc/src/main/asciidoc/tp2.adoc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -234,7 +234,7 @@ FeedDb Class:

[,java]
----
@Path("/ignite/tp2/feedDb")
@Path("/ignite-tp2/feedDb")
@Slf4j
@RequiredArgsConstructor
public class FeedDb {
Expand All @@ -243,7 +243,10 @@ public class FeedDb {
@SuppressWarnings("BigDecimalMethodWithoutRoundingCalled")
@GET
@Path("/{nbAccounts}/{nbAccountsInError}/{nbOperationsPerAccount}")
public Response feedTheDb(@PathParam("nbAccounts") Integer nbAccounts, @PathParam("nbAccountsInError") Integer nbAccountsInError, @PathParam("nbOperationsPerAccount") Integer nbOperationsPerAccount) {
public Response feedTheDb(
@DefaultValue("20") @PathParam("nbAccounts") Integer nbAccounts,
@DefaultValue("5") @PathParam("nbAccountsInError") Integer nbAccountsInError,
@DefaultValue("20") @PathParam("nbOperationsPerAccount") Integer nbOperationsPerAccount) {
ClientCache<AccountKey, Account> accounts = igniteClient.getOrCreateCache("accounts");
ClientCache<OperationKey, Operation> operations = igniteClient.getOrCreateCache("operations");
log.info("Get caches {} , {}",accounts,operations);
Expand Down
60 changes: 38 additions & 22 deletions doc/src/main/asciidoc/tp3.adoc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,17 +2,20 @@

== Why?

We have previously seen that it is quite simple to use the Java thin client to access Ignite. However, this connection mode has some limitations:
We have previously seen that it is quite simple to use the Java thin client to access Ignite.
However, this connection mode has some limitations:

* Fewer technical possibilities:
** The Ignite services we want to call must already be deployed on the Ignite server nodes.
** The client cannot really receive events or provide additional services to the grid.
** The DataStreamer, which allows for fast data feeding, is not available in the Thin Client.
* It forces us to treat Ignite servers as off-the-shelf products. This complicates delivery and operation procedures in some complex structures.
* It forces us to treat Ignite servers as off-the-shelf products.
This complicates delivery and operation procedures in some complex structures.

An Ignite node can be configured in "client mode", meaning it will not host data (except for near cache) but can "contribute" its services to the grid.

In this TP, given the limited time available, we cannot explore all the possibilities of the framework. That's why we have chosen to introduce you to some aspects that, in our opinion, are useful in a microservice-oriented architecture with Quarkus.
In this TP, given the limited time available, we cannot explore all the possibilities of the framework.
That's why we have chosen to introduce you to some aspects that, in our opinion, are useful in a microservice-oriented architecture with Quarkus.

== Presentation of the modules.

Expand All @@ -23,7 +26,8 @@ In TP3, we have 3 submodules:
** We will launch 2 instances of this microservice.
* node-service: a module that allows us to share common interfaces and models between "client" and "server."

In the client and server modules, the Boot and IgniteProducer classes are already ready to use. In addition to TP1 and TP2, besides the fact that we produce an 'Ignite' instance instead of 'IgniteClient', you will see some configuration differences:
In the client and server modules, the Boot and IgniteProducer classes are already ready to use.
In addition to TP1 and TP2, besides the fact that we produce an 'Ignite' instance instead of 'IgniteClient', you will see some configuration differences:

[,java]
----
Expand Down Expand Up @@ -51,7 +55,7 @@ igniteConfiguration.setClientMode(true);

First of all, let's compile the modules (if it's not already done):

[,shell]
[source,shell]
----
./mvnw clean install
----
Expand Down Expand Up @@ -109,7 +113,8 @@ The Swagger UIs are accessible here:

== Distributed Computing (https://ignite.apache.org/docs/latest/distributed-computing/distributed-computing)

This part of the API allows you to send code to execute on remote nodes. Very powerful, you can:
This part of the API allows you to send code to execute on remote nodes.
Very powerful, you can:

* Send code to execute on all nodes or according to a selection.
* Take advantage of load balancing and/or failover mechanisms.
Expand All @@ -124,17 +129,20 @@ To call "Pure Quarkus" services, whose classes are "managed," we recommend using
CDI.current().select(MyService.class).get().....
----

Also, be very careful with Anonymous classes, which, during serialization, carry the entire cluster of parent classes with them. This can sometimes cause issues during deserialization.
Also, be very careful with Anonymous classes, which, during serialization, carry the entire cluster of parent classes with them.
This can sometimes cause issues during deserialization.

=== Sending code to all selected nodes.

____
CAUTION: The code sent to a remote node will be executed in an Apache Ignite thread. If you want to use a Quarkus service, it may be necessary to use the @ContextActivator annotation on it.
CAUTION: The code sent to a remote node will be executed in an Apache Ignite thread.
If you want to use a Quarkus service, it may be necessary to use the @ContextActivator annotation on it.
____

*Create a REST service in the node-client that injects an Ignite instance; we will use it.*

Let's say we want to execute a task on all server-type nodes, such as clearing caches, compressing directories, etc. The Ignite compute API allows us to send code to the chosen nodes.
Let's say we want to execute a task on all server-type nodes, such as clearing caches, compressing directories, etc.
The Ignite compute API allows us to send code to the chosen nodes.

To do this, we use the *IgniteCompute* class and its *broadcast()* method.

Expand All @@ -158,7 +166,8 @@ public class ExerciceRest_3_3 {

(the 'ignite' variable is injected by Quarkus; the instance is created by the Producer).

However, before the compute method addresses the entire cluster, including client-mode nodes, it can take a ClusterGroup as a parameter that selects the target nodes. To describe a ClusterGroup, simply use:
However, before the compute method addresses the entire cluster, including client-mode nodes, it can take a ClusterGroup as a parameter that selects the target nodes.
To describe a ClusterGroup, simply use:

[,java]
----
Expand Down Expand Up @@ -207,9 +216,12 @@ You have probably noticed that this code uses remote class loading.

*Replace 'broadcast' with 'run,' then press F5 on the client's Swagger UI (with Quarkus, this is enough to recompile/deploy your app), and you will see that when you call your REST service, only one node receives the message.*

By default, the broadcast method sends to all nodes that meet the cluster selection. The other methods perform load balancing. By executing your REST call several times, you will observe the distribution of log messages in the two server consoles.
By default, the broadcast method sends to all nodes that meet the cluster selection.
The other methods perform load balancing.
By executing your REST call several times, you will observe the distribution of log messages in the two server consoles.

By default, jobs are evenly distributed in RoundRobin fashion. However, it is possible to configure *JobStealing* to distribute the load on unoccupied nodes (https://ignite.apache.org/docs/latest/distributed-computing/load-balancing).
By default, jobs are evenly distributed in RoundRobin fashion.
However, it is possible to configure *JobStealing* to distribute the load on unoccupied nodes (https://ignite.apache.org/docs/latest/distributed-computing/load-balancing).

*Injected Resources*

Expand Down Expand Up @@ -268,7 +280,8 @@ public Response anonymousClass() {

*Calls can also be asynchronous*

And you can consume the response reactively! Add a REST method to test this:
And you can consume the response reactively!
Add a REST method to test this:

[,java]
----
Expand All @@ -282,10 +295,13 @@ public Response async() {
}
----

The return code here is executed in a thread managed by Ignite. However, you can choose to use your own ExecutorService with the listenAsync method, allowing you to use a managed thread pool:
The return code here is executed in a thread managed by Ignite.
However, you can choose to use your own ExecutorService with the listenAsync method, allowing you to use a managed thread pool:

[,java]
----
private final ExecutorService managedExecutor = Executors.newSingleThreadExecutor();

@GET
@Path("listenAsyncManagedExecutor")
public void listenAsyncManagedExecutor() throws InterruptedException {
Expand All @@ -296,13 +312,14 @@ public void listenAsyncManagedExecutor() throws InterruptedException {
after -> log.info("Return handled in thread {}", Thread.currentThread()),
managedExecutor
);
Thread.sleep(20000); // this guarantees that the current thread will not be used when the response comes
Thread.sleep(10*1000); // this guarantees that the current thread will not be used when the response comes
}
----

*Jobs can also be assigned by affinity key on a cache*

There are other job execution APIs, such as map/reduce (https://ignite.apache.org/docs/latest/distributed-computing/map-reduce). You can attempt an implementation of the example in Ignite's documentation (ComputeTaskExample).
There are other job execution APIs, such as map/reduce (https://ignite.apache.org/docs/latest/distributed-computing/map-reduce).
You can attempt an implementation of the example in Ignite's documentation (ComputeTaskExample).

== Ignite Services (https://ignite.apache.org/docs/latest/services/services)

Expand Down Expand Up @@ -378,15 +395,14 @@ ignite.services(grp).deploy(

Then, to test on the node-client side, create a REST service:

[,java]
[source,java]
----
@GET
@Path("/callRemoteAdd/{a}/{b}")
public void callRemoteAdd(@PathParam("a")int a, @PathParam("b")int b){
MySimpleService mySimpleService = ignite.services(ignite.cluster().forAttribute("nodeType", "server"))
.serviceProxy("MySimpleService", MySimpleService.class, false);
log.info("Remote Call {} plus {} = {}", a, b, mySimpleService.add(a, b));

public void callRemoteAdd(@DefaultValue("20") @PathParam("a")int a, @DefaultValue("20") @PathParam("b")int b){
MySimpleService mySimpleService = ignite.services(ignite.cluster().forAttribute("nodeType", "server"))
.serviceProxy("MySimpleService", MySimpleService.class, false);
log.info("Remote Call {} plus {} = {}", a, b, mySimpleService.add(a, b));
}
----

Expand Down
6 changes: 0 additions & 6 deletions tp3/node-client/src/main/java/fr/sciam/lab/ignite/tp3/client/rest/ExerciceRest_3_4.java
View file
Open in desktop

This file was deleted.

13 changes: 0 additions & 13 deletions tp3/node-server/src/main/java/fr/sciam/lab/ignite/tp3/server/Boot.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,22 +19,9 @@
@RequiredArgsConstructor
@Slf4j
public class Boot {

public static final String RANDOM_INTEGERS = "randomIntegers";
private final Ignite ignite;

public void onBoot(@Observes StartupEvent startupEvent) {
log.info("Boot, create a Map of Random Integers");
IgniteCache<UUID, Integer> randomIntegers = ignite.getOrCreateCache(RANDOM_INTEGERS);
Random random = new Random();
AtomicInteger sum=new AtomicInteger(0);
for ( int index=0 ; index < 1000 ; index++) {
int val = random.nextInt(500);
sum.addAndGet(val);
randomIntegers.put(UUID.randomUUID(), val);
}
log.info("Sum is equal to {}",sum.get());

ClusterGroup grp = ignite.cluster().forAttribute("nodeType", "server");
log.info("grp {}",grp.nodes());
ignite.services(grp).deploy(
Expand Down