Swan Versions Save

v0.1.11

biggest achievement of this release is that we totally rewrite the state machine of Swan. For long time till now we have the issue of inflexible state machine, which stop us from being do rolling creation/scaling up/scaling down/rolling update sequentially.

With this release we are proud to say rolling update is true rolling update now, which seriously depends on how the health check was described. Only those tasks which successfully pass health check could marked as finished and proceed with next one.

The additional good that this feature bring is any scaling up/scaling down is preemptive now which means if scaling up is in progress, we can scale down or delete the app which stop the scaling up process, no need to wait it done before take next action.

[BUG] task IP lost after Swan restart [BUG] DNS A record lost after Swan restart for fixed app [IMPROVE] Handle recind offer [IMPROVE] Handle inverse offer [FEATURE] restart policy revisit [IMPROVE] remove mixed mode [IMPROVE] supprt host mode network for replicates app

[IMPROVE] rename appID to appName in version JSON, import appName which is the name of an app. [IMPROVE] make sure fixed type app allow customized their network name. [IMPROVE] make sure framework handle mesos leader shift gracefully [FEATURE] add node status and heartbeats from leader to agent

• [IMPROVEMENT] Constraints improvement, for more info please visit constraints
• [FEATURE] Agent can be removed out from the cluster
• [BUG] DNS record issue for fixed type application
• [IMPROVEMENT] ./bin/swan default without any arguments should display help
• [IMPROVEMENT] add help API to return all API access points of Swan
• [IMPROVEMENT] separate irrelevant components out of swan codebase. eg. swan cli, go swan, swan frontend(make it standalone swan-ui project)

we present framework as another mesos scheduler the same as sched. we want outdate sched in next release but now you have the option to switch them with flag with-engine.