hazelcast/hazelcast

Open Source In-Memory Data Grid

Sagar Raut

@v8sagar

Config config = new Config();
config.setLiteMember(false);
// setting up the local grid server
config.getNetworkConfig().setPublicAddress("0.0.0.0").setPort(8099).setPortAutoIncrement(true);

    // setting up the cluster
    System.out.println( Arrays.asList(environment).toString());
    JoinConfig join = config.getNetworkConfig().getJoin();
    join.getMulticastConfig().setEnabled(false);
    join.getAwsConfig().setEnabled(false);
    join.getTcpIpConfig().setEnabled(true).setMembers(Arrays.asList(environment));

    CacheSimpleConfig cacheConfig = new CacheSimpleConfig();
    cacheConfig.setName("buckets");
    config.addCacheConfig(cacheConfig);```

Jaromir Hamala

@jerrinot

setPublicAddress("0.0.0.0") looks wrong.

what environment is it? AWS?

Sagar Raut

@v8sagar

My approach I will run the two instance without creating the IP for HAZELCAST on the same server with the port number

then I will join the cluster with IP and PORT

Jaromir Hamala

@jerrinot

so there won't be anyone connecting from outside? only 2 Hazelcast instances running on the same server talking to each other?

Sagar Raut

@v8sagar

other will be connectin

Jaromir Hamala

@jerrinot

why do you set the public address at all? this usually needed only when you have someone in a different network connecting to the cluster.

usually the default public address should work fine. unless you are in a complicated environment such as Kuberentes. but for that you should use a Kubernetes plugin.

Sagar Raut

@v8sagar

I need to as I am going to join the cluster from a different server

Jaromir Hamala

@jerrinot

is the different server in the same network?

Sagar Raut

@v8sagar

same netrwok

Jaromir Hamala

@jerrinot

ok. in this case you need to provide details about your environment and topology. clearly setPublicAddress("0.0.0.0") is wrong.
the question is: what should be there instead? (if anything at all)

here is what I know so far:

you have 2 Hazelcast instances running on 1 server
there is another instance running on a different service.
all instances are in the same network
all 3 instances are supposed to form a single cluster

is :point_up: correct?

Sagar Raut

@v8sagar

1> yes
2>yes
3>yes
4>yes

@jerrinot What I am doing wrong?

Jaromir Hamala

@jerrinot

what if you remove setPublicAddress("0.0.0.0") ?

Sagar Raut

@v8sagar

then I don't know my hazelcast ip as it by default run on
192.168.122.1:5701

Jaromir Hamala

@jerrinot

setPublicAddress() sets an IP address others should use to talk to this member. this is usually needed when you have multiple networks and what not.
for some reason you have 0.0.0.0 which CANNOT ever be used to talk to this member. The address 0.0.0.0 has a special meaning in IP networks.

Sagar Raut

@v8sagar

so if i use my actual address will it work?

setPublicAddress("127.0.0.1")?

It worked thanks for clarifying it :)

Jaromir Hamala

@jerrinot

you have 3 options:

use multicast discovery. this is the default. it's simple to use and has no external dependency. but it does not work on most cluster providers as clouds do not allow multicast traffic.
use static configuration. then you have to know IP address of all your members. I see you use join.getTcpIpConfig().setEnabled(true).setMembers(Arrays.asList(environment)); for this joiner you have to know all IP addresses anyway.
use a discovery plugin. this is the most popular option when deploying in clouds. we have discovery plugins for AWS, Azure, GCP, Kubernetes, Zookeepr, Eureka, etc..

setPublicAddress("127.0.0.1") will work only on a single box. again, remove that. most likely you dont need at all.

Sagar Raut

@v8sagar

okay thanks

Jaromir Hamala

@jerrinot

you are very welcome. Happy Hazelcasting!

Sagar Raut

@v8sagar

@jerrinot
for my case, this will work perfectly fine right?

    Config config = new Config();
    config.setLiteMember(false);
    JoinConfig join = config.getNetworkConfig().getJoin();
    config.getNetworkConfig().getJoin().getMulticastConfig().setEnabled(false);
    config.getNetworkConfig().getJoin().getAwsConfig().setEnabled(true)
            .setProperty("tag-key", "my-ec2-instance-tag-key")
            .setProperty("tag-value", "my-ec2-instance-tag-value");

    CacheSimpleConfig cacheConfig = new CacheSimpleConfig();
    cacheConfig.setName("buckets");
    config.addCacheConfig(cacheConfig);

    HazelcastInstance hazelcastInstance = Hazelcast.newHazelcastInstance(config);
    ICacheManager cacheManager = hazelcastInstance.getCacheManager();
    Cache<String, GridBucketState> cache = cacheManager.getCache("buckets");

    return cache;

1> 2 Hazelcast instances running on 1 server
2>there is another instance running on a different service.
3>all instances are in the same network
4>all 3 instances are supposed to form a single cluster

Jaromir Hamala

@jerrinot

that looks fine to me.
some notes: config.setLiteMember(false); is redundant as that's the default.

depending on your deployment scheme you might want to store hazelcastInstance somewhere and shut it down when your application is about to shutdown. this is mostly concern if you deploy to an app server with multiple tenants (applications). If you are using something like Spring Boot (or appserver-less deployment in general) then it's usually not a concern.

other than that - it looks good to me

if you are on AWS then using the AWS plugin for discovery is the best option

Sagar Raut

@v8sagar

yes I am on AWS

and I am using springboot

Sagar Raut

@v8sagar

@jerrinot if this works
I will be creating Github repo with lots of explanation for someone like me
Happy Hazelcasting

Jaromir Hamala

@jerrinot

that would be excellent! @mesutcelik would be certainly interested in that!

Sagar Raut

@v8sagar

@jerrinot I wanted to know one addidtional thing
if my all three instances are now same network
will this work

   Config config = new Config();
    config.setLiteMember(false);
    System.out.println( Arrays.asList(environment).toString());
    JoinConfig join = config.getNetworkConfig().getJoin();
    join.getMulticastConfig().setEnabled(false);
    join.getAwsConfig().setEnabled(false);
    join.getTcpIpConfig().setEnabled(true).setMembers(Arrays.asList("172.323.24.130","896.341.438.65"));

Hazelcast

@hazelcast_twitter

[Rafal Leszko (unknown)] @v8sagar If the IPs are correct, then you static TCP/IP configuration will work, but as mentioned, you're better off using AWS Discovery plugin when running on AWS.

Sagar Raut

@v8sagar

@hazelcast_twitter thank you

Vignesh-Thiraviam

@Vignesh-Thiraviam

Hi I need to disable some URL's in hazel cast management centre REST api's

Jaromir Hamala

@jerrinot

@Vignesh-Thiraviam what's your Hazelcast version?

Vignesh-Thiraviam

@Vignesh-Thiraviam

@jerrinot Hazelcast management centre ,version 3.12.8

Jaromir Hamala

@jerrinot

@emre-aydin please see :point_up: Is this possible at all?

Vignesh-Thiraviam

@Vignesh-Thiraviam

@jerrinot I have enabled Clustered rest in hazelcast management centre , but it provides a lot of API's , need to enable only MAP reads

Vignesh-Thiraviam

@Vignesh-Thiraviam

@hazelcast_twitter can you help on this

Sagar Raut

@v8sagar

hi all if set this config, and same code deployed on two different AWS instance
api-1 and api-2 both are talking to each other will this work

 config.getNetworkConfig().getJoin().getMulticastConfig().setEnabled(false);
    config.getNetworkConfig().getJoin().getAwsConfig().setEnabled(true)
            .setProperty("tag-key", "Name1")
            .setProperty("tag-value", " api-1")
            .setProperty("access-key","1")
            .setProperty("secret-key","1");

    config.getNetworkConfig().getJoin().getMulticastConfig().setEnabled(false);
    config.getNetworkConfig().getJoin().getAwsConfig().setEnabled(true)
            .setProperty("tag-key", "Name2")
            .setProperty("tag-value", " api-2")
            .setProperty("access-key","2")
            .setProperty("secret-key","2");

Hazelcast

@hazelcast_twitter

[Rafal Leszko (unknown)] Yes, it looks correct assuming you have tags configured on your EC2 Instances and the port 5701 is open in security groups

Sagar Raut

@v8sagar

Just to rephrase
same configs code have deployed on both the servers I mean api-1 and api-2

config.getNetworkConfig().getJoin().getMulticastConfig().setEnabled(false);
config.getNetworkConfig().getJoin().getAwsConfig().setEnabled(true)
.setProperty("tag-key", "Name1")
.setProperty("tag-value", " api-1")
.setProperty("access-key","1")
.setProperty("secret-key","1");

config.getNetworkConfig().getJoin().getMulticastConfig().setEnabled(false);
config.getNetworkConfig().getJoin().getAwsConfig().setEnabled(true)
        .setProperty("tag-key", "Name2")
        .setProperty("tag-value", " api-2")
        .setProperty("access-key","2")
        .setProperty("secret-key","2");

above code is constant

Hazelcast

@hazelcast_twitter

[Rafal Leszko (unknown)] yes, it's fine

Jaromir Hamala

@jerrinot

@Vignesh-Thiraviam the clustered API in Managmeent Centre is all-or-nothing. It does not allow fine grained per-resource configuration for now.

the new Management Centre for Hazelcast 4 has a Prometheus exporter and you will be able to define what to export.

Where communities thrive

People

Repo info

Activity