Eskimo Logo   Eskimo.sh Eskimo Components Download Eskimo Eskimo CE Documentation

Eskimo Community Edition
Release Notes

Read below the various Eskimo Version release notes

What's new in Eskimo-CE v0.3

Released on Jul 5, 2020

What's new ?

  • Integrated marathon. Individual services - so called unique services in Eskimo terminology are now operated on Mesos with the help of Mesosphere Marathon. This is a key step forward in scaling these services and implementing increasingly High Availability as well on the Community Version of Eskimo.
    Marathon is fully supported in operating services and managing their lifecycles and fully configured with dedicated configuration screens on Eskimo thanks to the Marathon Sub-system.
    Services operated by marathon are proxied by Eskimo and the user doesn't need to have any understanding or knowlegde of where they run at runtime.
  • Fixed Zeppelin ElasticSearch inerpreter preventing from upgrading ElasticSearch to version 7.X Fixed Zeppelin v0.9 ElasticSearch interpreter which was not compatible with ElasticSearch version 7.x or above. The fix is implemented as a eskimo custom distribution of Zeppelin based on Zeppelin 0.9-preview1.
  • Using gluster as shared FileSystem for marathon services. Leveraging on gluster for sharing folders between marathon services and services running on node host. Due the locality abstraction provided by Marathon, no assumption of operation can be performed on node.
    Gluster is now used even on single node deployments to share storage folders between marathon services and host services.
  • Fine tuning of mesos resources . Administrators can now fine-tune the resources declared by mesos-agents on eskimo cluster nodes. This has become an even more important requrements now that most services are move to marathon / mesos.
  • Utility command to encode passwords. Provided an utility command-line program called encode-password.sh/bat to generate the encoded password to be stored in eskimo users definition file users.json for new users.
  • Mesos monitoring dashboard in Grafana. Leveraging on Prometheus and Grafana, Mesos activity can now be monitored in Grafana, both for Mesos Agents and the Mesos Master. An initial dashboard is provisionned with Eskimo.
  • New custom command framework. Administrators integrating eskimo services can now configure custom commands available from the status page when clicking on a service status icon in the "Nodes Status Table".
    Eskimo pre-packaged services implement access to the software services log files by leveraging on this custom commands framework.
  • Eskimo SystemD Unit Configuration file. A SystemD Unit configuration file eskimo.service is added to the distribution package along with a script __install-eskimo-systemD-unit-file.sh aimed at installing the service on systemD with all system level installation and configuration required to enable it to run properly.
  • Base components version update.
    Switched from OpenJDK 8 to OpenJDK 11 for all services running on Java / Scala except Spark which supports only JDK 9 (as long as Spark 3 is not out in final version
    Switched from Stretch to Buster as base Debian Docker image.
  • Various improvements on the integrated SSH Console.
    • Using Ctrl + Shift + Left/Right to navigate back and forth between opened consoles to the various nodes
    • Supporting copy paste using keyboard shortcuts (see user guide)
    • Consoles are now sized automatically to the available window space.
  • General improvement and bug corrections.
    • Fixed the problem preventing Flink App Master and Marathon from running on a different node than the Mesos Master by configuring properly LIB_PROCESS (mesos native library).
    • Significant improvement of gluster shares management and operation reliability.
    • Significant improvements on services monitoring, detection and management.
    • Significant improvements on processed management and monitoring in containers running several processes.
    • Implemented a custom Eskimo solution do handle removing marathon services docker containers that mesos-agents fail to remove.
    • Fixed spark wrappers such as spark-shell and spark-sql that were unable to reach the mesos cluster.
    • Significantly improve services and nodes configuration consistency enforcement.
    • Provisioning a Kibana dashboard for berka transactions (Zeppelin samples)
  • Services Development framework improvement.
    • The Service development framework now leverages on docker as well to build the various Apache Mesos distributions. Vagrant or libvirt is not required anymore.
  • Services Version Update.
    • Elastic Stack (ElasticSearch, Logstash, Kibana) upgraded to 7.6.2
    • Apache Spark upgraded to 2.4.5
    • Apache Flink upgraded to 1.10.1
    • Cerebro updated to 0.9.2
    • Kafka Manager updated to 3.0.0.3
    • Apache Zeppelin upgraded to 0.9-preview-1 (some Eskimo workarounds for known bugs still required)
    • NTP updated to 4.2.8p12 (standard debian buster version)
    • GlusterFS updated to 6.0 (standard debian buster version)
    • Zookeeper updated to 3.4.13 (standard debian buster version)
    • Prometheus upgraded to 2.18.1
      • Pushgateway exporter updated to 1.2.0
    • Grafana upgraded to 6.7.4
  • Services Integration. Following services are fully integrated and operable within Eskimo 0.3, with all the tuning, fixes and required wrappers.
    • Mesosphere Marathon 1.8.222
    • Mesos Prometheus Exporter 1.1.2

What's new in Eskimo-CE v0.2

Released on Dec 30, 2019

What's new ?

  • SSH Tunnels to reach Services Web UIs. From the eskimo main console, instead of reaching the various Web Graphical User Interfaces of the various services such as Mesos Console, Spark History Server, Kafka Manager, Cerebro, Kibana, etc. using direct accesses to the nodes and ports where these services are running, the new approach consists in letting eskimo manage SSH tunnels to reach these services wherever they are located and let the iframes within Eskimo's own UI reach these services through these SSH tunnels using proper HTTP Proxying from the Eskimo Backend.
    This enables a much tighter control on the ports to be opened on individual nodes and prevents from the need to open access from administrators machines to the internal Eskimo cluster nodes.
    It also enables a much tighter integration between service web consoles and Eskimo itself.
  • Revamped Status Page. The Nodes Status Page is now a Cluster Status Page that provides general health information and statistics about the cluster the whole cluster. The services action menu has been reimplemented in a clearer way (just click on a service status to have access to the action menu)
  • Download of images from eskimo.sh. Service Images and Packages can be downloaded from eskimo.sh. It is not anymore mandatory to build them locally with each and every Eskimo installation. Images and Packages are properly versionned and the system supports checking for updates and fetching new images when they are available.
  • Support for Microsoft Windows as execution OS for Eskimo backend. With the ability to download pre-built services packages (docker images) from eskimo.sh now implemented, Microsoft Windows is now fully supported as execution middleware for the Eskimo backend (not the cluster nodes though, a supported Linux distribution is required on the eskimo cluster nodes).
    Running on windows however makes it mandatory to download pre-built packages from a remote repository (any one). Building images directly from within the Eskimo User Interface is only supported when the backend Operating System is Linux.
  • Logstash is available from zeppelin container. If Logstash is installed, it is made available to the Zeppelin container using a command client reaching a command server from the logstash container.
  • Mesos Management Command Line utility. A mesos-cli command line is available from every node of the cluster to administer the mesos cluster and address missing features from mesos command line utilities such as for instance the ability to kill a framework in a failsafe manner.
  • Suport for SUSE operation System. The SUSE Operating System is now fully supported on cluster nodes in addition to Red-Hat based OSes (Fedora, CentOS, etc.) and Debian based OSes (Debian, Ubuntu, etc.)
  • Collection of Zeppelin Demo Notebooks. A comprehensive collection of demo notebooks for Zeppelin is now available.
    These demo notebooks present for instance how to implement spark processes in batch and streaming mode, same for flink, how to read from and write to ElasticSearch from Spark, how to read from and write to kafka from Spark and Flink, how to use logstash to feed ElasticSearch, etc.
  • New Services Settings Edition Feature. It is now possible to edit common spark, flink, logstash, elasticsearch, kafka, etc. configuration properties or settings from the UI and inject them in the service runtime (services are automatically restarted with updated configuration when settings are saved).
    Only pre-defined configuration properties are supported in Eskimo Community Edition. One needs to acquire the enterprise edition to support defining every possible property or configuration file of every Eskimo pre-packages technology.
  • General improvement and bug corrections.
    • Much better management of SSH Connections and SSH terminals.
    • Much better support for different screen resolutions, especially lower screen resolutions.
    • Much better detection of services startup problems in SystemD unit configuration files.
    • Much better management of gluster shares. (A weakness remains on node unregistration when a node is removed from cluster. This will be solved in 0.3)
    • Better handling of messaging and notifications in a multi-user environment.
    • Improvement of the user documentation.
  • Services Development framework improvement.
    • The Service development framework now supports both libvirt and Vagrant to build the various Apache Mesos distributions.
  • General UI Improvements.
    • The services in both the Nodes Status Page and the Services Configuration Page are now accompanied with the respective product icons.
  • Services Version Update.
    • Apache Mesos upgraded to version 1.8.1
    • Apache Zeppelin upgraded to 0.9-SNAPSHOT (development version with fixes and workarounds for known bugs)
  • Services Integration. Following services are fully integrated and operable within Eskimo 0.2, with all the tuning, fixes and required wrappers.
    • Apache Flink 1.9.1

What's new in Eskimo-CE v0.1

Released on Jul 24, 2019

What's new ?

  • Docker Images Development Framework. The docker images development framework provides standards, principles and tools to enable anyone to build his own service docker images to be installed and operated by Eskimo. The initial set of Eskimo packaged services containers are implemented on top of this framework.
  • Services Installation Framework. The Services Installation framework is Eskimo's key feature enabling it to install, configure, manage, operate, move and uninstall services (their docker containers) on node. It is based on docker and a standard systemd approach for operation on nodes.
  • Master election and dependencies Management. Dependencies management between services and enforcement of dependencies between services accross nodes. This enables an eskimo service developer to define it's service dependencies in a configuration file and then let eskimo enforce these dependencies, handle order of installation, order of services restart, restarting of dependent services after a service move to another node, etc.
    This module also takes care of dependent services re-configuration when topology of dependencies evolve.
  • Initial Basic Memory Management Framework. This module computes the available memory on every node and uses the memory requirement declaration of every service (configuration) to distribute a fair share of the available memory on every node to every service the node is hosting.
  • Eskimo Platform Management User Interface. This is the intial version of the Eskimo Platform Management Console Web Graphical User Interface which provides specific features, such as the SSH terminals, the nodes configuration page, etc. as well as the embdedding of third party services User Interfaces within its own UI.
  • Eskimo Setup Page and Backend. The Eskimo Setup Page and Backend implements the initial setup of the Eskimo Platform management console, where the SSH configuration used to reach the Eskimo Cluster nodes is defined as well as, for instance, the way to find the Eskimo Managed Services Docker images (download or build).
  • Eskimo Nodes Configuration Page and Backend. This is the most important feature from eskimo: the page and backend implementation of the nodes configuration, where the topolog of the cluster is defined by declaring the services to be executed on every node of the cluster. Nodes can be configured as individual nodes or range of nodes (IP addresses). This module takes care controls the dependencies enforcement and performes the installation, configuration, uninstallation, etc. of services when the configuration is applied.
  • Eskimo Nodes Status Page< and Backend. This is the Eskimo Cluster main monitoring page where the topology of the cluster is presented and monitored and where individual services states can be monitored. Actions on individual services can be carried on from there.
  • Eskimo Backend Messages Page and Backend. Messaging and monitoring feature when there are operations in progress (installation, configuration, uninstallation, etc.) on the backend. This included support for multi-user environment when another administrator triggered some operations to have all connected operators notified.
  • Eskimo Web-based SSH Terminals Feature. Eskimo SSH terminals feature, just as a plain old SSH terminal but within a web page. Eskimo enables administrators to open SSH terminals on cluster nodes directly from within the Eskimo Web GUI.
  • Eskimo Web-based SFTP Terminals Features. Web based file manager using SFTP to connect on cluster nodes. Just as a plain old file manager with file visualization but from within the Eskimo Web GUI.
  • Support for Debian-based and Red-Hat-based Linux nodes. Support for Debian-based and Red-Hat-based Operating systems on Eskimo cluster nodes (Debian, Ubuntu, Red-Hat, Fedora and CentOS for now).
  • Basic authentication and authorization and login page. Authentication and Authorization framework. For now authorization are quite trivial and authentication is based on local file system user file. This shall evolve with further versions of Eskimo.
    Eskimo EE offers impersonation of system users when executing commands (includign File Manager and SSH terminal). Eskimo CE executes everything as default user.
  • Gluster FS volumes and mounts management framework. System-wide management of Gluster shares and mounts for services requiring such shared folders such as Spark, Zeppelin, etc.
  • Services Integration. Following services are fully integrated and operable within Eskimo 0.1, with all the tuning, fixes and required wrappers.
    • NTP (standard debian stretch version)
    • GlusterFS (standard debian stretch version)
    • Zookeeper (standard debian stretch version)
    • Apache Mesos version 1.7.1
    • Apache Spark version 2.4.4
    • ElasticSearch version 6.8.3
    • Elastic Logstash version 6.8.3
    • Elastic Kibana version 6.8.3
    • Zeppelin version 0.8.1
    • Cerebro version 0.8.4
    • Apache kafka version 2.2.0
    • Kafka Manager version 2.0.0.1
    • Grafana version 6.3.3
    • Gdash (Gluster Dashboard) version 0.0.a1
    • Prometheus version 2.10.0 with
      • Node Exporter version 0.18.1
      • Push Gateway version 0.8.0