mesos 0.2.10

Mesos library using the new HTTP API failed to build mesos-0.2.10
Please check the build logs for more information.
See Builds for ideas on how to fix a failed build, or Metadata for how to configure builds.
If you believe this is' fault, open an issue.
Visit the last successful build: mesos-0.2.9

mesos-rs :globe_with_meridians:

Simple bindings for the Mesos v1 HTTP API. The default trait of SchedulerRouter routes to a traditional mesos callback interface, but you are free to provide your own router implementation for working with other coding styles if you prefer. Flexibility is the primary concern with this library.


  • scheduler
  • executor
  • zk master detection and failover


mesos = "0.2.10"
extern crate mesos;

use self::mesos::{Scheduler, SchedulerClient, SchedulerConf,
            ProtobufCallbackRouter, run_protobuf_scheduler};
use self::mesos::proto::*;
use self::mesos::util;

struct TestScheduler {
    max_id: u64,

impl TestScheduler {
    fn get_id(&mut self) -> u64 {
        self.max_id += 1;

impl Scheduler for TestScheduler {
    fn subscribed(&mut self,
                  client: &SchedulerClient,
                  framework_id: &FrameworkID,
                  heartbeat_interval_seconds: Option<f64>) {
        println!("received subscribed");


    // Inverse offers are only available with the HTTP API
    // and are great for doing things like triggering
    // replication with stateful services before the agent
    // goes down for maintenance.
    fn inverse_offers(&mut self,
                      client: &SchedulerClient,
                      inverse_offers: Vec<&InverseOffer>) {
        println!("received inverse offers");

        // this never lets go willingly
        let offer_ids = inverse_offers.iter()
                                      .map(|o| o.get_id().clone())
        client.decline(offer_ids, None);

    fn offers(&mut self, client: &SchedulerClient, offers: Vec<&Offer>) {

        // Offers are guaranteed to be for the same agent, and
        // there will be at least one.
        let agent_id = offers[0].get_agent_id();

        println!("received {} offers from agent {}",

        let offer_ids: Vec<OfferID> = offers.iter()
                                            .map(|o| o.get_id().clone())

        // get resources with whatever filters you need
        let mut offer_cpus: f64 = offers.iter()
                                        .flat_map(|o| o.get_resources())
                                        .filter(|r| r.get_name() == "cpus")
                                        .map(|c| c.get_scalar())
                                        .fold(0f64, |acc, cpu_res| {
                                            acc + cpu_res.get_value()

        // or use this if you don't require special filtering
        let mut offer_mem = util::get_scalar_resource_sum("mem", offers);

        let mut tasks = vec![];
        while offer_cpus >= 1f64 && offer_mem >= 128f64 {
            let name = &*format!("sleepy-{}", self.get_id());

            let task_id = util::task_id(name);

            let mut command = CommandInfo::new();
            command.set_value("env && sleep 10".to_string());

            let mem = util::scalar("mem", "*", 128f64);
            let cpus = util::scalar("cpus", "*", 1f64);

            let resources = vec![mem, cpus];

            let task_info = util::task_info(name,
            offer_cpus -= 1f64;
            offer_mem -= 128f64;
        client.launch(offer_ids, tasks, None);

    fn rescind(&mut self, client: &SchedulerClient, offer_id: &OfferID) {
        println!("received rescind");

    fn update(&mut self, client: &SchedulerClient, status: &TaskStatus) {
        println!("received update {:?} from {}",

    fn message(&mut self,
               client: &SchedulerClient,
               agent_id: &AgentID,
               executor_id: &ExecutorID,
               data: Vec<u8>) {
        println!("received message");

    fn failure(&mut self,
               client: &SchedulerClient,
               agent_id: Option<&AgentID>,
               executor_id: Option<&ExecutorID>,
               status: Option<i32>) {
        println!("received failure");

    fn error(&mut self, client: &SchedulerClient, message: String) {
        println!("received error");

    fn heartbeat(&mut self, client: &SchedulerClient) {
        println!("received heartbeat");

    fn disconnected(&mut self) {
        println!("disconnected from scheduler");

fn main() {
    let mut scheduler = TestScheduler { max_id: 0 };

    let conf = SchedulerConf {
        master_url: "http://localhost:5050".to_string(),
        user: "root".to_string(),
        name: "rust http".to_string(),
        framework_timeout: 0f64,
        implicit_acknowledgements: true,
        framework_id: None,

    // If you don't like the callback approach, you can implement
    // an event router of your own.  This is merely provided for
    // those familiar with the mesos libraries in other languages.
    let mut router = ProtobufCallbackRouter {
        scheduler: &mut scheduler,
        conf: conf.clone(),

    run_protobuf_scheduler(&mut router, conf)