Examples with JobStatus com.spotify.helios.common.descriptors.JobStatus used on opensource projects

Example 16 with JobStatus

use of com.spotify.helios.common.descriptors.JobStatus in project helios by spotify.

the class ZooKeeperHeliosFailoverTest method undeploy.

private void undeploy(final JobId jobId) throws Exception {
    // Check job status can be queried
    final JobStatus jobStatus = client.jobStatus(jobId).get();
    assertEquals(RUNNING, jobStatus.getTaskStatuses().get(testHost()).getState());
    // Undeploy the job
    final JobUndeployResponse undeployed = client.undeploy(jobId, testHost()).get();
    assertEquals(JobUndeployResponse.Status.OK, undeployed.getStatus());
    // Wait for the task to disappear
    awaitTaskGone(client, testHost(), jobId, LONG_WAIT_SECONDS, SECONDS);
}

Example 17 with JobStatus

use of com.spotify.helios.common.descriptors.JobStatus in project helios by spotify.

the class ExpiredJobReaper method runOneIteration.

@Override
public void runOneIteration() {
    for (final Entry<JobId, Job> entry : masterModel.getJobs().entrySet()) {
        final JobId jobId = entry.getKey();
        final Job job = entry.getValue();
        if (job.getExpires() == null) {
            //noinspection UnnecessaryContinue
            continue;
        } else if (job.getExpires().getTime() <= clock.now().getMillis()) {
            final JobStatus status = masterModel.getJobStatus(jobId);
            final List<String> hosts = ImmutableList.copyOf(status.getDeployments().keySet());
            for (final String host : hosts) {
                try {
                    masterModel.undeployJob(host, jobId, job.getToken());
                } catch (HostNotFoundException e) {
                    log.error("couldn't undeploy job {} from host {} when it hit deadline", jobId, host, e);
                } catch (JobNotDeployedException e) {
                    log.debug("job {} was already undeployed when it hit deadline", jobId, e);
                } catch (TokenVerificationException e) {
                    log.error("couldn't undeploy job {} from host {} because token verification failed", jobId, host, e);
                }
            }
            try {
                masterModel.removeJob(jobId, job.getToken());
            } catch (JobDoesNotExistException e) {
                log.debug("job {} was already removed when it hit deadline", jobId, e);
            } catch (JobStillDeployedException e) {
                log.debug("job {} still deployed on some host(s) after expiry reap", jobId, e);
            } catch (TokenVerificationException e) {
                log.error("couldn't remove job {} because token verification failed", jobId, e);
            }
        }
    }
}

Example 18 with JobStatus

use of com.spotify.helios.common.descriptors.JobStatus in project helios by spotify.

the class SystemTestBase method undeployJob.

protected void undeployJob(final JobId jobId, final String host) throws Exception {
    final String undeployOutput = cli("undeploy", jobId.toString(), host);
    assertThat(undeployOutput, containsString(host + ": done"));
    final String output = cli("status", "--host", host, "--json");
    final Map<JobId, JobStatus> statuses = Json.readUnchecked(output, new TypeReference<Map<JobId, JobStatus>>() {
    });
    final JobStatus status = statuses.get(jobId);
    assertTrue(status == null || status.getDeployments().get(host) == null);
}

Example 19 with JobStatus

use of com.spotify.helios.common.descriptors.JobStatus in project helios by spotify.

the class FlappingTest method test.

@Test
public void test() throws Exception {
    // CircleCI boxes are too slow -- the job doesn't stop or restart fast enough to ever flap
    assumeFalse(isCircleCi());
    startDefaultMaster();
    final String host = testHost();
    startDefaultAgent(host);
    final HeliosClient client = defaultClient();
    awaitHostStatus(client, host, UP, LONG_WAIT_SECONDS, SECONDS);
    final Job flapper = Job.newBuilder().setName(testJobName).setVersion(testJobVersion).setImage(BUSYBOX).setCommand(asList("nc", "-p", "4711", "-l")).addPort("poke", PortMapping.of(4711)).build();
    final JobId jobId = createJob(flapper);
    deployJob(jobId, host);
    awaitTaskState(jobId, host, RUNNING);
    // Poke the container to make it exit until it's classified as flapping
    Polling.await(LONG_WAIT_SECONDS, SECONDS, new Callable<Object>() {

        @Override
        public Object call() throws Exception {
            final JobStatus jobStatus = getOrNull(client.jobStatus(jobId));
            final TaskStatus taskStatus = jobStatus.getTaskStatuses().get(host);
            if (taskStatus.getThrottled() == FLAPPING) {
                return true;
            }
            final PortMapping port = taskStatus.getPorts().get("poke");
            assert port.getExternalPort() != null;
            poke(port.getExternalPort());
            return null;
        }
    });
    // Verify that the job recovers after we stop poking
    awaitJobThrottle(client, host, jobId, NO, LONG_WAIT_SECONDS, SECONDS);
}

Example 20 with JobStatus

use of com.spotify.helios.common.descriptors.JobStatus in project helios by spotify.

the class HealthCheckTest method testContainerDiesDuringHealthcheck.

@Test
public void testContainerDiesDuringHealthcheck() throws Exception {
    startDefaultMaster();
    final HeliosClient client = defaultClient();
    startDefaultAgent(testHost(), "--service-registry=" + registryAddress);
    awaitHostStatus(client, testHost(), UP, LONG_WAIT_SECONDS, SECONDS);
    final HealthCheck healthCheck = TcpHealthCheck.of("health");
    final Job job = pokeJob(healthCheck);
    final JobId jobId = createJob(job);
    deployJob(jobId, testHost());
    awaitTaskState(jobId, testHost(), HEALTHCHECKING);
    // kill the underlying container
    final JobStatus jobStatus = getOrNull(client.jobStatus(jobId));
    final TaskStatus taskStatus = jobStatus.getTaskStatuses().get(testHost());
    getNewDockerClient().killContainer(taskStatus.getContainerId());
    // ensure the job is marked as failed
    final int timeout = WAIT_TIMEOUT_SECONDS;
    Polling.await(timeout, SECONDS, new Callable<Object>() {

        @Override
        public Object call() throws Exception {
            final TaskStatusEvents jobHistory = getOrNull(client.jobHistory(jobId));
            for (final TaskStatusEvent event : jobHistory.getEvents()) {
                if (event.getStatus().getState() == FAILED) {
                    return true;
                }
            }
            return null;
        }
    });
    // wait for the job to come back up and start healthchecking again
    awaitTaskState(jobId, testHost(), HEALTHCHECKING);
    pokeAndVerifyRegistration(client, jobId, timeout);
}