marathon-0.7.6
On behalf of the Marathon contributors, we are pleased to announce
the availability of version 0.7.6.
Overview:
- Added GROUP_BY constraint for hostname.
- An event is emitted when Marathon registers with, re-registers, or disconnects from Mesos.
- Task checkpointing is now enabled by default.
- Added a per-app parameter for Docker privileged mode.
- Added a per-app parameter to pass arbitrary options to
docker run
. - Added handling of new Mesos task state TASK_ERROR.
- The timeout for I/O operations to the persistence layer is now configurable.
- Began migrating JSON (de)serialization to play-json.
- Task names are now set to the fully-qualified app ID (instead of the task ID).
- Added a restart app primitive to simplify rolling deploy implementation.
- Overhauled deployment plan calculation, yielding less stages and actions in general.
- More Mesos and Marathon metadata is available in the task environment (app ID, task ID, slave ID).
- Marathon now periodically queries Mesos for the list of all known tasks via reconciliation.
- The default task reconciliation interval has been increased from 30 seconds to 5 minutes.
- The run script passes JVM options in JAVA_OPTS to Marathon.
- Various bugfixes and improvements.
Changelog from Marathon 0.7.5 to 0.7.6:
v0.7.5...v0.7.6
Tarball:
http://downloads.mesosphere.io/marathon/v0.7.6/marathon-0.7.6.tgz
SHA:
http://downloads.mesosphere.io/marathon/v0.7.6/marathon-0.7.6.tgz.sha256
CentOS/RHEL6:
http://downloads.mesosphere.io/marathon/v0.7.6/marathon-0.7.6-1.0.x86_64.rpm
Debian/Ubuntu 14.04:
http://downloads.mesosphere.io/marathon/v0.7.6/marathon_0.7.6-1.0_amd64.deb