2018 Releases

v3.46 - 23 February 2018

This release contains a few foundational features have been added (external objects, Docker connector, credential sharing) that will improve cloud resource management in the future. It also includes changes to the way machines within a deployment access the server and how deployment reports are stored. Both require administrator attention during upgrades. (See migration section.) The release also contains a number of bug fixes.

For Everyone:
  • User resource implementation was changed to allow credential sharing between users and groups with ACLs.
  • The login dialog was changed to avoid it being obscured on mobile devices.
  • The default ACL for Connector resources was changed to allow all authentication users to see them.
  • The bootstrap script has been corrected to avoid an issue where machine deployments on Ubuntu 16 machines would fail.
  • The prototype for the new web browser UI has been improved to provide better editing capabilities with forms and JSON, to plot server metrics, and to render href attributes as links to other resources.
  • Styles of cubic (new web browser UI) have been normalized to provide a consistent look and feel.
For Clara:
  • The login methods of the Python API have been improved to cache credentials to make managing access easier.
  • Improved the CIMI support in the Python API to allow CIMI actions to be called.
  • The Python API is now part of the SlipStream RPM packages.
  • A utility method was added to the Python API to retrieve deployment events.
  • A function was added to the Clojure(Script) API to allow the server metrics to be retrieved.
  • A prototype “cloud” connector (alpha) for Docker infrastructures is now available.
For Dave:
  • The “machine” cookies that were used by VMs within a deployment to interact with SlipStream have been replaced by an API key/secret pair. These can be revoked if necessary.
  • An “external object” CIMI resource has been created to allow links to external files and resources, such as report, data files, etc. Reports are now handled with these resources. (See migration below.)
  • The server organization has been more finely segmented to allow for wider reuse of the servers and to make containerization easier.
  • Package dependencies have be rationalized and corrected (including the cheshire.jar verson in the pricing service). More work on this will occur in the future to reduce the servers’ footprints.
  • SlipStream package dependency on slipstream-client-clojure (no longer created) has been removed.

Alice, Bob, Clara, and Dave can be found here.

Migration

API key/secret pairs are now being used to manage access to the server from deployed machines. For non-scalable deployments, this change will have no effect. However, scalable deployments will lose access to the server. They need to be terminated and restarted.

Below is the migration procedure to enable the view of the connector instances by users of your SlipStream instance. From now on this is required for the deployments to succeed.

  • login to SlipStream instance as super user

  • go to https://<slipstream>/webui/cimi/connector

  • click on magnifying glass pictogram (this will fetch all connector config instances)

  • click on a connector name link

  • click on update button

  • in the edit window add the following into the list under “acl” -> “rules”:

    { "principal": "USER", "right": "VIEW", "type": "ROLE" }
    
  • click on update button to persist the configuration

  • repeat this for each connector.

The method of storing reports has changed with this release. They are now stored in S3 rather than on the server’s disk. This requires that the administrator have access to an S3 instance and migration of the existing reports to S3.

You must provide an S3 configuration file /opt/slipstream/server/.aws/credentials with the following contents:

aws_secret_access_key=<KEY>
aws_access_key_id=<SECRET>
aws_endpoint=<S3ENDPOINT>

Note that the name of the bucket is not configurable. It is set to “slipstream-reports” and must be created before being used.

Known issues

  • The switch to using API key/secret pairs will only have an effect on running scalable deployments. These will need to be stopped and redeployed.

v3.45 - 4 February 2018

This is primarily a bug fix release, but also includes a prototype for a new web interface. Feedback on that prototype is welcome.

For Everyone:
  • An SSH configuration bug that blocked SSH logins on machines without pre-existing .ssh directories was fixed.
  • A bug with the Exoscale connector that caused deployments to fail was corrected.
  • A prototype user interface has been included in the release, which is available by default on the /webui relative URL.
For Dave:
  • The configuration for the Job Engine has been added to the quick installation script.
  • CIMI resources for NuvlaBox registrations have been added.
  • Unnecessary dependencies have been removed from services and packages have been cleaned up.

Alice, Bob, Clara, and Dave can be found here.

Migration

No migration is necessary.

Known issues

  • When upgrading rename the /etc/default/ssclj file to /etc/default/cimi if you’ve made changes to the configuration file.
  • If you’ve made changes to the nginx configuration files, you will need to remove the reference to authn.block in /etc/nginx/conf.d/slipstream.params.
  • The wrong version of cheshire.jar was included in the RPM package for the ss-pricing service. Replace /opt/slipstream/ss-pricing/lib/cheshire.jar with version 5.8.0 that can be found at clojars.org.
  • The RPM package slipstream-client-clojure was not generated for this release. The v3.44 version works fine.

v3.44 - 24 January 2018

This is primarily a bug fix release that makes improvements for SlipStream administrators.

For Everyone:
  • Fix bug in the deployment garbage collection that caused the clean up to fail.
For Dave:
  • Extend OpenNebula and NuvlaBox connectors to allow the vCPU/CPU ratio to be defined. The default value is 0.5.
  • Correct Logstash and Filebeat configurations when using the installation script to avoid having Logstash logs fill with errors.

Alice, Bob, Clara, and Dave can be found here.

Migration

No migration is necessary.

Known issues

No known issues.

v3.43 - 22 January 2018

For Everyone:
  • Remove deprecated basic authentication and related parameters from the Python API and Command Line Client.
  • Fix concurrency issue with cookie handling in the Python API and Command Line Client.
For Dave:
  • Mark /etc/default/slipstream as a configuration file to avoid having the configuration overwritten on upgrades.
  • Improve template handling for the Exoscale connector so that the most recent templates are used by default. Avoids a problem with running in the DK region.
  • Rationalize and reduce the default logging of the services to allow for better discovery and debugging of problems.
  • Optimize the loading of the user resources from the database and provide metrics.
  • Minimize connections requests from connectors on initialization.
  • Refactor the deployment garbage collector to reduce object creation and database churn.

Alice, Bob, Clara, and Dave can be found here.

Migration

No migration is necessary.

Known issues

  • Missing dependency for the ssclj server blocks start of service.

v3.42 - 12 January 2018

This release brings the following changes.

The persistence of the user entities was moved from HSQLDB to Elasticsearch and from now on will be managed through CIMI server. On already deployed systems this assumes that a migration is required. Check Migration section below.

The following connectors were removed and are no longer supported

  • StratusLab
  • Azure
  • NativeSoftlayer
  • VCloud

Fixes and improvements:

  • fixed and improved VMs information collection service.

Migration

The steps below perform migration of users from HSQLDB to Elasticsearch (via CIMI server).

Download migration script:

$ wget https://raw.githubusercontent.com/slipstream/SlipStreamServer/master/rpm/src/main/migrations/020_migrate_users_to_cimi.py
$ chmod +x 020_migrate_users_to_cimi.py
$ yum install python-lxml
$ # or
$ pip install lxml

Dump users with:

$ export SLIPSTREAM_USERNAME=super
$ export SLIPSTREAM_PASSWORD=<password>
$ ss-login --endpoint https://<slipstream>
$ ./020_migrate_users_to_cimi.py --endpoint https://<slipstream> --get users-3.41/

Perform the upgrade:

$ yum upgrade -y
$ systemctl restart hsqldb ss-pricing ssclj slipstream \
     [email protected]_collect \
     [email protected]_cleanup \
     [email protected]_cleanup \
     slipstream-job-executor \
     elasticsearch logstash filebeat kibana

In https://<slipstream>/configuration -> SlipStream Basics -> java class names remove any instances of the following connectors: nativesoftlayer, stratuslab, stratuslabiter, azure, vcloud. Save the configuration.

Push users back to SlipStream:

$ ./020_migrate_users_to_cimi.py --endpoint https://<slipstream> --put users-3.41/

Known issues

  • No known issues.