diff options
| author | Fabian Mastenbroek <mail.fabianm@gmail.com> | 2022-05-06 17:47:44 +0200 |
|---|---|---|
| committer | Fabian Mastenbroek <mail.fabianm@gmail.com> | 2022-05-06 17:47:44 +0200 |
| commit | a9657e4fa3b15e2c1c11884b5a250b0861bcc21d (patch) | |
| tree | 6b25de3d7a1def150ab4977a45723c52167e7211 /opendc-web | |
| parent | 48da4538707cd074969287724ca6f02823f2ff5a (diff) | |
| parent | 8e3905273c7a3f2df4df5d5840e4088d99b0dffb (diff) | |
merge: Expose metrics directly to user (#80)
This pull request adds the ability to access the metrics of resources modeled
by the OpenDC Compute, Workflow, FaaS, and TensorFlow services directly from
their corresponding interfaces. Previously, users would have to interact with
OpenTelemetry to obtain these values, which is complex and provides
significant overhead.
With this pull request, users can access the metrics of all cloud resources
modeled by OpenDC via methods such as `getSchedulerStats()`, etc.
** Breaking Changes **
- `ComputeService.hostCount` removed in favour of `ComputeService.hosts.size`
Diffstat (limited to 'opendc-web')
| -rw-r--r-- | opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/OpenDCRunner.kt | 29 | ||||
| -rw-r--r-- | opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/internal/JobManager.kt | 2 | ||||
| -rw-r--r-- | opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/internal/WebComputeMonitor.kt (renamed from opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/internal/WebComputeMetricExporter.kt) | 3 |
3 files changed, 16 insertions, 18 deletions
diff --git a/opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/OpenDCRunner.kt b/opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/OpenDCRunner.kt index a150de4e..7c0c43ed 100644 --- a/opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/OpenDCRunner.kt +++ b/opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/OpenDCRunner.kt @@ -23,8 +23,9 @@ package org.opendc.web.runner import mu.KotlinLogging +import org.opendc.compute.api.Server import org.opendc.compute.workload.* -import org.opendc.compute.workload.telemetry.SdkTelemetryManager +import org.opendc.compute.workload.telemetry.NoopTelemetryManager import org.opendc.compute.workload.topology.HostSpec import org.opendc.compute.workload.topology.Topology import org.opendc.compute.workload.topology.apply @@ -35,13 +36,12 @@ import org.opendc.simulator.compute.model.ProcessingUnit import org.opendc.simulator.compute.power.LinearPowerModel import org.opendc.simulator.compute.power.SimplePowerDriver import org.opendc.simulator.core.runBlockingSimulation -import org.opendc.telemetry.compute.collectServiceMetrics -import org.opendc.telemetry.sdk.metrics.export.CoroutineMetricReader +import org.opendc.telemetry.compute.ComputeMetricReader import org.opendc.web.client.runner.OpenDCRunnerClient import org.opendc.web.proto.runner.Job import org.opendc.web.proto.runner.Scenario import org.opendc.web.runner.internal.JobManager -import org.opendc.web.runner.internal.WebComputeMetricExporter +import org.opendc.web.runner.internal.WebComputeMonitor import java.io.File import java.time.Duration import java.util.* @@ -180,9 +180,9 @@ public class OpenDCRunner( private val scenario: Scenario, private val repeat: Int, private val topology: Topology, - ) : RecursiveTask<WebComputeMetricExporter.Results>() { - override fun compute(): WebComputeMetricExporter.Results { - val exporter = WebComputeMetricExporter() + ) : RecursiveTask<WebComputeMonitor.Results>() { + override fun compute(): WebComputeMonitor.Results { + val monitor = WebComputeMonitor() // Schedule task that interrupts the simulation if it runs for too long. val currentThread = Thread.currentThread() @@ -206,25 +206,24 @@ public class OpenDCRunner( else null - val telemetry = SdkTelemetryManager(clock) val simulator = ComputeServiceHelper( coroutineContext, clock, - telemetry, + NoopTelemetryManager(), computeScheduler, failureModel, interferenceModel.takeIf { phenomena.interference } ) - - telemetry.registerMetricReader(CoroutineMetricReader(this, exporter, exportInterval = Duration.ofHours(1))) + val servers = mutableListOf<Server>() + val reader = ComputeMetricReader(this, clock, simulator.service, servers, monitor) try { // Instantiate the topology onto the simulator simulator.apply(topology) // Run workload trace - simulator.run(vms, seeder.nextLong()) + simulator.run(vms, seeder.nextLong(), servers) - val serviceMetrics = collectServiceMetrics(telemetry.metricProducer) + val serviceMetrics = simulator.service.getSchedulerStats() logger.debug { "Scheduler " + "Success=${serviceMetrics.attemptsSuccess} " + @@ -235,14 +234,14 @@ public class OpenDCRunner( } } finally { simulator.close() - telemetry.close() + reader.close() } } } finally { interruptTask.cancel(false) } - return exporter.collectResults() + return monitor.collectResults() } } diff --git a/opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/internal/JobManager.kt b/opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/internal/JobManager.kt index 8de0cee4..99b8aaf1 100644 --- a/opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/internal/JobManager.kt +++ b/opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/internal/JobManager.kt @@ -62,7 +62,7 @@ internal class JobManager(private val client: OpenDCRunnerClient) { /** * Persist the specified results. */ - fun finish(id: Long, results: List<WebComputeMetricExporter.Results>) { + fun finish(id: Long, results: List<WebComputeMonitor.Results>) { client.jobs.update( id, Job.Update( diff --git a/opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/internal/WebComputeMetricExporter.kt b/opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/internal/WebComputeMonitor.kt index 04437a5f..69350d8c 100644 --- a/opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/internal/WebComputeMetricExporter.kt +++ b/opendc-web/opendc-web-runner/src/main/kotlin/org/opendc/web/runner/internal/WebComputeMonitor.kt @@ -22,7 +22,6 @@ package org.opendc.web.runner.internal -import org.opendc.telemetry.compute.ComputeMetricExporter import org.opendc.telemetry.compute.ComputeMonitor import org.opendc.telemetry.compute.table.HostTableReader import org.opendc.telemetry.compute.table.ServiceTableReader @@ -32,7 +31,7 @@ import kotlin.math.roundToLong /** * A [ComputeMonitor] that tracks the aggregate metrics for each repeat. */ -internal class WebComputeMetricExporter : ComputeMetricExporter() { +internal class WebComputeMonitor : ComputeMonitor { override fun record(reader: HostTableReader) { val slices = reader.downtime / SLICE_LENGTH |
