v1.3.0

We're glad to announce the Fogbow v1.3.0!

The main features coming out in this release are:

• Implementation of accounting endpoint and admin user's profile, in this way an admin user may get information about accounting from the fogbow manager, such as: user's usage; how much resource was consumed/donated from/to other members.
• Supporting federation of Storage resource. The interface of StoragePlugin was defined and currently it's possible to federate compute and storage resouces. For now, we are providing three implementation for it: OpenStackV2StoragePlugin; OpenNebulaStorragePlugin; and CloudStackStoragePlugin.
• Implementation of creation/remotion of attachment. If the compute instance and the storage volume were created in the same member, it's possible attach the volume to the instance and, once it is attached, it's possible detach as well.
• Implementation of load balance of Fogbow Reverse Tunneling. Currenlty the maximum number of SSH servers and their connections are configured and a new SSH server is started or stoped on demand.

And bug fixes and improvements from previous version. Major changes are:

• Renaming of requests to orders, beside of renaming the user needs to specify which kind of resource he is ordering (compute or storage).
• Addition of Usage and Quota information in the member endpoint. In this way the user may retrieve information about his usage and/or quota in a specific fogbow manager.
• Implementation of keeping information about private IP of the compute instance, previously the only network information kept was the SSH public IP.

v1.2.1

The Fogbow team is glad to announce the Fogbow Manager v1.2.1

This release comes with bug fixes and improvements from v1.2.0. Major changes are:

• Use of SQLite instead of H2 as database technology due a bug in the H2 that makes database file always increasing;
• Creation of a new message between managers in order to remove served requests by requesting member. It is possible that a request to be forwarded to more than one member, when it is FULFILLED by someone of these members, the requesting member should delete requests have already sent to other members related to that one FULFILLED now;
• Remotion of resource capacity information from exchanged messages between Fogbow Rendezvous and Fogbow Manager. Rendezvous does not have these information anymore and the managers must exchange message with each other to figure out capacity related to the current federation user;
• Remotion of configuration about "I Am Alive" message period. This property is set by Rendezvous component from now and may be updated dynamically;
• Renaming of Local Credential plugin to Mapper plugin;
• Implementation of PENDING state of request. It means that request is not FULFILLED yet, but it was forwarded to a federation member;
• Fix of bug in RoundRobinMemberPicker plugin when allowed member list was empty, it was returning the last member picked instead of none.

And new plugins implementations, such as:

• FedeationUserMapperPlugin
• VOWhiteListAuthorizationPlugin

v1.2.0

We're glad to announce the Fogbow v1.2.0!

The main features coming out in this release are:

• Remotion of support of local cloud token;
• Implementation of the generic abstraction of mapping federation users in underlying cloud users as LocalCredentialsPlugin. At this version, we are providing three implementation for it: Single, VOBased, and MemberBased;
• Persistence of requests, allowing Fogbow Manager to recover previous state when it is restarted;
• Improvement of request information while getting request description (e.g. requesting and providing members);
• Implementation of creation of compute instance directly, the user now may create federated instances calling a post method at compute endpoint, then this is mapped to a request at Fogbow Manager. These compute instances are persisted and recovered when Fogbow Manager is restarted.

v1.1.0

The Fogbow team gladly releases Fogbow v1.1.0!

This release comes with a number of new features, such as:

• Implementation of Cloudstack Compute and Identity plugins;
• Implementation of EC2 Compute and Identity plugins;
• Implementation of Azure Compute and Identity plugins;
• Implementation of Shibboleth Identity Plugin, which enables interoperability with CAFe;

And bug fixes and improvements from v1.0.0. Major changes are:

• Deprecation of OCCI bypass calls. This feature will not be maintained and will not be available in the next releases. The Fogbow Manager will only support OCCI extensions that enables federated operations, such as the ones related to requests;
• Performance improvements related to request scheduling by trying to avoid unnecessary communication with the underlying cloud, e.g. when user did not pass its own token and yet we tried to create an instance with an invalid token;
• Update of HTTPClient to version 4.4.1 due to bugs in the previous version;
• Creation of the Fogbow maven repository on github that replaces the local repository copy inside each Fogbow related project.

v1.0.0

The Fogbow team gladly jumps a major digit by releasing Fogbow v1.0.0!

Besides a number of refactoring and bug fixes from v0.3.2, v1.0.0 comes with the following new features:

• Users can now provide an extra cloud-init user-data file within an instance request at the Fogbow Manager.
• Additional tunnels created at the Fogbow Reverse Tunnelling service can now be queried as instances' attributes.
• An executable called create-fogbow-tunnel is now injected on instances created by the Fogbow Manager and makes the creation of additional tunnels for services besides SSH much simpler.
• When a local user is logged, the Fogbow Dashboard will now display her quota at the Members' view.

For more issues addressed in this version, please visit: https://github.com/fogbow/fogbow-manager/issues?q=milestone%3Av1.0.0+is%3Aclosed.

Enjoy it!

v0.3.2

We're glad to announce the Fogbow Manager v0.3.2!

The major changes coming out in this release are:

• Forced prioritization of local or remote requests in cases of preemption. Such behaviour can be configured in the configuration file.
• Image Storage plugin for VMCatcher. When creating a Fogbow request, users can now provide an URL to an image.list that VMCatcher will subscribe to.
• Benchmarking plugin for SSH remote execution. When new instances become available to a Fogbow Manager, this plugin will execute a configurable command via SSH and time this execution. This metric will be used in other to derive the instance's computing power and eventually be used by the Accounting plugin when updating members' usage.

You can find all issues addressed in this release at https://github.com/fogbow/fogbow-manager/issues?q=milestone%3Av0.3.2+is%3Aclosed.

Enjoy it

v0.3.1

We're glad to announce the Fogbow Manager v0.3.1!

In this release we have merged in two major features: prioritization with NoF and match making via requirements.

Implementations of the Prioritization plugin will be invoked by the scheduler when in contention, which, in its turn, will tell the Manager whether there should be a preemption and which instance to be preempted. The NoF implementation tries to improve fairness by balancing what each Manager donates to and receives from other sites of the federation.

Users can now specify requirements in terms of RAM and vCPU instead of flavours when creating requests. That mitigates the administrative overhead that comes with the association between global flavours and local flavours in configuration, which may lead to unrealistic or outdated associations between those two abstractions.

Besides those, a number of issues have been addressed in this last iteration.

Enjoy it!

v0.3.0

The Fogbow team is glad to announce the Fogbow Manager v0.3.0

Besides bug fixing, the main features coming out in this release are:

• Resource usage accounting. Managers now save historical usage data by local users and remote members.
• Implementations of MemberValidator now receive the requesting User credentials as a paramenter. This is particularly useful for a federated identity approach with VOMS.
• Managers now ping the green sitter when no resources are available locally. The green sitter, on its turn, will try to wake up hosts that it previously put to sleep.

v0.2.3

We're glad to announce the Fogbow Manager v0.2.3!

This release comes with a series of fixes for major issues, such as:

• VOMS plugin failing to parse proxy certificates with line breaks
• Local user token being used in requests where the federation user token should be used instead
• Cloud-init not working for VMs spawned by OpenNebula
• A request forwarded to a remote member would go stale if the respective reply never arrives

Also, new features were implemented:

• Requests now hold two tokens: a federation level token, that grants the creation of the request itself at the Fogbow Manager; and a local cloud token, that is passed to the cloud when new instances are requested. This feature allows users to use their local quota while being authenticated in a federated IdP, e.g. VOMS
• Orphan instances - i.e. instances created by the Fogbow user and no longer associated to any request, due to the Fogbow Manager getting restarted, for example - are periodically collected and removed
• Images can now be specified with absolute URLs
• OVA support

Enjoy it!

v0.2.2

We are glad to announce the Fogbow Manager v0.2.2!

Fogbow is an umbrella for cloud-related projects, that, together, enable a federated and oppostunistic cloud.

The primary focus of this release was to make the Fogbow Manager a fully OCCI compliant server.

Besides fixes related to OCCI and bug fixes in general, we have adopted global identifiers for instances, following the format instance_id@member_id. This fix avoids clashes of remote instances' identifiers.

We have also changed the way remote requests are performed, moving from a synchronous and serial to an asynchronous and parallel fashion. Now multiple remote requests can be performed at the same time - making allocation much faster.

This release also comes with the support for dynamic upload of images fetched from VM marketplaces to the underlying cloud. That means that requests for images that are not present in the local cloud will trigger queries for images in public image repositories, and, eventually, these images will become available locally.

Enjoy it!