by

Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
    Hazelcast
    @hazelcast_twitter
    [Rafal Leszko (unknown)] you can ssh into the POD and check the mounted path permissions
    Andreas Höhmann
    @ahoehma
    ah .. okay ... cool :D I will check this
    another question ... I can't ssh into the "mancenter-0" because its not running :-) I ssh into the hazelcast-N pods .. and there is no "/data" mount (I guess this is right because they don't need it) ... sry for my noob questions :)
    ähmm... update ... the "/data" is also in the pods :)
    image.png
    Andreas Höhmann
    @ahoehma
    Yuhu :D ... the docker update fixed my problem ! :D
    Hazelcast
    @hazelcast_twitter
    [Rafal Leszko (unknown)] :)
    silverj
    @silverj
    Hey, I noticed that hazelcast-zookeeper has a 4.0 release in github but it is not available in public maven repositories. Are you aware of any plans of getting it published there?
    Grzegorz Piwowarek
    @pivovarit
    @silverj if it's not there, it's a bug. Will be fixed as soon as possible
    Yes, definitely a bug. It should be there in a few hours
    silverj
    @silverj
    Thanks for the quick response
    Grzegorz Piwowarek
    @pivovarit
    Fixed. However, we needed to make a new release (4.0.1) since the deployment was caused by an error in one of the source files
    Mohammed Faizuddin
    @FaizuddinEverest

    Hi everyone, I'm trying to put a simple docker container running on an EC2 instance without Auto Discovery. I'm trying to connect with a client but am getting a connection timed out error. I'm confident I opened the right ports in the security group. I'm also sure I put the right container ports and host ports. I also set the local.publicAddress to the instance public IP.

    Is there some other configuration that I should set?

    2 replies
    Andreas Höhmann
    @ahoehma
    Hello ... does anyone of you have an idea how to run hazelcast for (client-client-communication - sharing data) and a other caching-framework (ehcache) for real "client-local/client-only" caching? It seems that springboot can (per default) only handle one of both :(
    Vassilis Bekiaris
    @vbekiaris
    @ahoehma I don't know how to handle this case with spring-boot, but since you mentioned local caching, have you taken a look at Hazelcast near cache? You can configure your client to maintain a local cached copy of values. You can specify an eviction policy to determine how near cache reacts to reaching size limits and there 's a ton of other useful features like the preloader to get your near cache quickly warm after a restart.
    Andreas Höhmann
    @ahoehma
    @vbekiaris thanks for the hint ... but with "local-cache" I really mean an area where only "this" client caches some stuff ... all the other services in my "cluster" should not known these data ... u know ... would this be possible with hazelcast? Can I define a cache "client-only" or something?
    Vassilis Bekiaris
    @vbekiaris
    I see, near cache is always backed by a cluster data structure (IMap , Cache etc), so it's always a layer on top of shared data
    Andreas Höhmann
    @ahoehma
    May I can ask a more general question ... could be that my current idea how to use hazelcast is not right :) I would like to create my own "cluster-software". I would like to have a springboot-ms ... can handle some rest-api ... then I want to run many of them (k8s) and each of them got some settings what kind of "products" he can create and so on. In addition all the created products should be available (readonly) for ALL the cluster-members. So my idea was to use here a hazelcast shared-cache. Performance is a big topic ... do u think hazelcast is a good choice?
    Vassilis Bekiaris
    @vbekiaris
    A shared Hazelcast cache sounds like a good fit for this use case. When you are using embedded Hazelcast members, apart from near cache + IMap you also have the choice of using ReplicatedMap so all reads will be local (if your dataset fits in a single member's memory).
    Andreas Höhmann
    @ahoehma
    The values of my map are "only" json-documents, the keys are "uuids" ... simple start :) The json-documents are quite big ... may I will compress the strings and put this to the map ... or is there some "auto-compress-feature" in hazelcast?
    silverj
    @silverj
    Inconsistency in Hazelcast docs - this part suggests using hazelvast-client version 4.0.1: https://docs.hazelcast.org/docs/latest/manual/html-single/#connecting-clients-to-remote-member but migration guide for 4.0 says hazelcast-client no longer exists. One of them is wrong, I suspect migration guide is the correct one
    Serdar Ozmen
    @Serdaro
    True @silverj , I will check
    Thanks a lot for the notice
    Andreas Höhmann
    @ahoehma
    Make it sense to use hazelcast to implement a service-registry? I played with eureka or even k8s service-registry and I think there is one thing missing ... the posibilty to select a service based on "some rules". In my case I would like to route customer requests to services which can "really handle" such a request (product A is possible on service A, product B is possible on service B and C, ...). And here the IMDG comes in my mind to share a "service-discovery-map" between all the cluster-members (serviceId->collection of products). Would be nice to get some feedback ... maybe someone already implemented that? :D
    Mesut Celik
    @mesutcelik
    @ahoehma Can you please describe the "rules" part? If they are business rules then maybe a rules engine would help you on that. See https://www.drools.org/
    If it is more traffic re-routing for canary deployment or distribution the load between stateless services then service meshes(i.e Istio) can help you out.
    2 replies
    Marcos F. Lobo
    @marcosflobo
    Hi all, I have a question about hits on specific values stored in a cache map
    I'd like to know if it's possible to get the number of hits for an especific key in a cache map
    Ioannis Christodoulou
    @JChrist
    I don't seem to be able to find out how to configure a node to start (perhaps like fresh/first-member?) when existing cluster node(s) fail to respond to join requests (e.g. crashed/unresponsive). I'm trying to simulate this behavior using a "frozen" single node cluster (frozen using a debugger and hitting a breakpoint) and then trying to start a new node (in new jvm)
    I've tried to change configuration values for GroupProperty.MAX_JOIN_SECONDS and GroupProperty.MASTERSHIP_CLAIM_TIMEOUT_SECONDS, but that results in the new node failing faster, not claiming cluster mastership and starting
    Hazelcast
    @hazelcast_twitter
    [Sharath Sahadevan (Sharath Sahadevan)] @marcosflobo - Entry view ( https://docs.hazelcast.org/docs/latest/javadoc/com/hazelcast/core/EntryView.html ) should give you the hits . You can find an example in the reference manual ( section 7.2.11). Hope that helps
    Ioannis Christodoulou
    @JChrist
    I tried to get into more details about my "issue". I believe that the new node does not "blacklist" the existing one, due to the socket getting connected and there is currently no way for a "joining" node to decide to blacklist an existing using no-response/no-heartbeat information. Thus it gets to an "eternal" loop in TcpIpJoiner:
    • tryJoinAddresses(possibleAddresses); -> existing node will not respond, thus it will not join
    • tryClaimMastership(possibleAddresses); -> existing node will not respond, thus it will not claim mastership
    As things stand, only if it eventually blacklisted the misbehaving node it would claim itself as the new master (via clusterJoinManager.setThisMemberAsMaster(); but this will never happen if the underlying socket gets established but nothing else gets through
    Am I (remotely) right? And more importantly is there some workaround for this scenario?
    Can Gencer
    @cangencer
    We've set up a slack workspace for Hazelcast Users, which you can join here: https://hz-community-slack.herokuapp.com/ (right now we're experimenting with it, so bear with us..)
    Marcos F. Lobo
    @marcosflobo
    @hazelcast_twitter yes, that solves my question, thanks!
    David Brimley
    @dbrimley
    Hi All. We’re running a survey with the Hazelcast Community and we’d like to find out what features you use and importantly, what new features would you like to see? If you get value from Hazelcast would you please support us by filling out this brief 10 question survey. It should take no longer than a few minutes….
    https://www.surveymonkey.co.uk/r/NQ6KF5X
    Many thanks and happy hazelcasting!
    Matko Medenjak
    @mmedenjak
    For those who haven't joined already, I highly encourage to join the community slack - https://hz-community-slack.herokuapp.com/. It's in the early stages but already we have stuff like announcements, separate channels for IMDG, jet and cloud, place where you can ask questions for contributing and we're hoping on expanding on it even further.
    jacob goldverg
    @jgoldverg
    Hi everyone, quick question. I was looking at the AWS marketplace, and was wondering if there is any intention of having Hazelcast available as a managed service??
    Hazelcast
    @hazelcast_twitter
    [Mike Yawn (unknown)] Yes, a managed service is available today on AWS, and coming soon for other clouds. https://hazelcast.com/products/cloud/
    3 replies
    Hazelcast
    @hazelcast_twitter

    [Rafal Leszko (unknown)] @jgoldverg ECS/Fargate support is released in hazelcast-aws:3.2 [1], but it's not included in the latest Docker image yet. So you need to either:

    1. Wait for the release of Hazelcast 4.0.2 (it will be included in the docker image hazelcast/hazelcast:4.0.2 OR
    2. Build your own image including hazlecast-aws:3.2 dependency [2]

    If you use Hazelcast embedded, is should just work when you include the dependency hazelcast-aws:3.2. Also, when we release Hazelcast 4.0.2, I'll write a blog post and Code Sample on how to set up Hazelcast on Fargate

    [1] https://github.com/hazelcast/hazelcast-aws/releases/tag/v3.2
    [2] https://github.com/hazelcast/hazelcast-docker#building-your-own-hazelcast-image

    1 reply
    ghj1040110333
    @ghj1040110333
    /**
     * After {@link Node#shutdown(boolean)} call completes, node's state will be {@code SHUT_DOWN}.
     * In {@code SHUT_DOWN} state node will be completely inactive. All operations/invocations
     * will be rejected. Once a node is shutdown, it cannot be restarted.
     */
    SHUT_DOWN
    What is the meaning of it?
    client  error :notifyError(new HazelcastInstanceNotActiveException("State: " + state + " Operation: " + op.getClass()));
    ghj1040110333
    @ghj1040110333
    Once a node is shutdown, it cannot be restarted.
    I now let the node restart but report the above error
    ???
    hazelcast 3.11.4
    Andreas Höhmann
    @ahoehma
    What can I do if my hazelcast client registration via eureka hangs? com.hazelcast.spi.discovery.integration.DiscoveryService: [192.168.2.107]:5701 [dev] [3.12.7] Waiting for registration with Eureka...
    ....
    Jaromir Hamala
    @jerrinot
    hi @ahoehma a threadump could help. this will show where is it hanging.
    Andreas Höhmann
    @ahoehma
    I found something ... now afterwards sounds logic to me ... I have to enabled "fetch-registry" :) without that hazelcast can't ask eureka for all the other.