Tweaked vSwarm deployer

.gitattributes

@@ -0,0 +1 @@
+workloads/container/vSwarm_deploy_metadata.tar.gz filter=lfs diff=lfs merge=lfs -text

.github/workflows/e2e_loader.yaml

@@ -39,7 +39,7 @@ jobs:
         run: go run cmd/loader.go --config pkg/config/test_config.json
 
       - name: Check the output
-        run: test -f "data/out/experiment_duration_5.csv" && test $(cat data/out/experiment_duration_5.csv | wc -l) -gt 1 && test $(grep true data/out/experiment_duration_5.csv | wc -l) -eq 0 # test the output file for errors (true means failure to invoke)
+        run: test -f "data/out/experiment_duration_2.csv" && test $(cat data/out/experiment_duration_2.csv | wc -l) -gt 1 && test $(grep true data/out/experiment_duration_2.csv | wc -l) -eq 0 # test the output file for errors (true means failure to invoke)
 
       - name: Print logs
         if: ${{ always() }}
@@ -54,4 +54,31 @@ jobs:
       - name: Down
         if: ${{ always() }}
         run: |
-          kn service delete --all
+          kn service delete --all
+
+      - name: Untar vSwarm YAMLs
+        if: ${{ always() }}
+        run: |
+          tar -xzvf workloads/container/vSwarm_deploy_metadata.tar.gz -C workloads/container/ 
+      - name: Run vSwarm loader
+        run: go run cmd/loader.go --config pkg/config/test_vswarm_config.json
+
+      - name: Check vSwarm output
+        run: test -f "data/out/experiment_duration_2.csv" && test $(cat data/out/experiment_duration_2.csv | wc -l) -gt 1 && test $(grep true data/out/experiment_duration_2.csv | wc -l) -le 1 # test the output file for errors (true means failure to invoke)
+
+      - name: Print logs
+        if: ${{ always() }}
+        run: |
+          set -x
+          container_list=$(kubectl get pods -n default -o jsonpath="{.items[*].spec.containers[*].name}")
+          for container_name in $container_list
+          do
+            kubectl logs -n default -c $container_name -l serving.knative.dev/service=${{ matrix.service }}
+          done
+      - name: Down
+        if: ${{ always() }}
+        run: |
+          kn service delete --all
+
+
+

.gitignore

@@ -15,6 +15,7 @@ pkg/workload/openwhisk/*.zip
 tools/*/__pycache__
 tools/trace_synthesizer/*.csv
 tools/trace_synthesizer/*/*.csv
+workloads/container/yamls/
 
 ### CMake ###
 CMakeLists.txt.user
@@ -209,4 +210,4 @@ tools/plotter/test-out
 *.swp
 
 data/traces/azure_*
-data/traces/day*
+data/traces/day*

cmd/config_vswarm_trace.json

@@ -0,0 +1,32 @@
+{
+  "Seed": 42,
+
+  "Platform": "Knative",
+  "InvokeProtocol" : "grpc",
+  "YAMLSelector": "container",
+  "EndpointPort": 80,
+
+  "BusyLoopOnSandboxStartup": false,
+
+  "TracePath": "data/traces/example/",
+  "Granularity": "minute",
+  "OutputPathPrefix": "data/out/experiment",
+  "IATDistribution": "exponential",
+  "CPULimit": "1vCPU",
+  "ExperimentDuration": 10,
+  "WarmupDuration": 0,
+
+  "IsPartiallyPanic": false,
+  "EnableZipkinTracing": false,
+  "EnableMetricsScrapping": false,
+  "MetricScrapingPeriodSeconds": 15,
+  "AutoscalingMetric": "concurrency",
+
+  "GRPCConnectionTimeoutSeconds": 15,
+  "GRPCFunctionTimeoutSeconds": 900,
+  "DAGMode": false,
+  "EnableDAGDataset": true,
+  "Width": 2,
+  "Depth": 2,
+  "VSwarm": true
+}

cmd/loader.go

@@ -173,11 +173,17 @@ func parseTraceGranularity(cfg *config.LoaderConfiguration) common.TraceGranular
 func runTraceMode(cfg *config.LoaderConfiguration, readIATFromFile bool, writeIATsToFile bool) {
 	durationToParse := determineDurationToParse(cfg.ExperimentDuration, cfg.WarmupDuration)
 	yamlPath := parseYAMLSpecification(cfg)
+	var functions []*common.Function
+	var traceParser trace.Parser
 
 	// Azure trace parsing
-	traceParser := trace.NewAzureParser(cfg.TracePath, durationToParse)
-	functions := traceParser.Parse()
+	if !cfg.VSwarm {
+		traceParser = trace.NewAzureParser(cfg.TracePath, durationToParse, yamlPath)
+	} else {
+		traceParser = trace.NewMapperParser(cfg.TracePath, durationToParse)
+	}
 
+	functions = traceParser.Parse()
 	// Dirigent metadata parsing
 	dirigentMetadataParser := trace.NewDirigentMetadataParser(cfg.TracePath, functions, yamlPath, cfg.Platform)
 	dirigentMetadataParser.Parse()

data/traces/example/mapper_output.json

@@ -0,0 +1,14 @@
+{
+    "c13acdc7567b225971cef2416a3a2b03c8a4d8d154df48afe75834e2f5c59ddfc455703077a17a9b8d0fc655d939fcc6d24d819fa9a1066b74f710c35a43cbc868baea05aa0c3619b6feb78c80a07e27e4e68f921d714b8125f916c3b3370bf2": {
+        "proxy-function": "cartservice"
+    },
+    "a2faad786b3c813b12ce57d349d5e62f6d0f22ceecfa86cd72a962853383b600d7028b01f2422ea9d4f695cae4085800eb34540674081a46bb4e4388e7995f7ac8b8b9160f181010f509ee0af3266cbcb5a5f4c7700ba8d4396f1147e1ad3bbd": {
+        "proxy-function": "image-rotate-go-11"
+    },
+    "7dc5aeabc131669912e8c793c8925cc9928321f45f13a4af031592b4611630d70728f480194febeddbc0d2a69a2cb38344e495f264d52c62102e7bd99505dbf22cc1c836a3b32265b4a36b6e6f56b2d7c0588a926df5d03a76b7b4388cbf2258": {
+        "proxy-function": "video-processing-python-100"
+    },
+    "ae8a1640fa932024f59b38a0b001808b5c64612bd60c6f3eb80ba9461ba2d09101d1a13d091b0cfc764a125ead054ad6a720cb29b0b1fc2c187e9be3311f09fe90ba6ab2b1b9f5b5925d1e9ee6242d0f840ebcfb3221567bc2b5f24cb864b016": {
+        "proxy-function": "video-processing-python-10"
+    }
+}

docs/configuration.md

@@ -31,6 +31,7 @@
 | EnableDAGDataset             | bool      | true/false                                                          | true                | Generate width and depth from dag_structure.csv in TracePath[^8]                                                                                                                                                                         |
 | Width                        | int       | > 0                                                                 | 2                   | Default width of DAG                                                                                                                                                                                                                     |
 | Depth                        | int       | > 0                                                                 | 2                   | Default depth of DAG                                                                                                                                                                                                                     |
+| VSwarm                       | bool      | true/false                                                          | false               | Execute vSwarm functions from mapper_output.json                               |
 
 [^1]: To run RPS experiments replace the path with `RPS`.
 

docs/generate_deploy_info_docs.md

@@ -0,0 +1,40 @@
+# Deploy Info JSON
+
+The `deploy_info.json` file is used to identify deployment information for services and their pre-dependencies. It is pre-generated, and stored inside the `vSwarm_deploy_metadata.tar.gz` to contain deployment information for vSwarm functions.
+
+## Schema and Usage
+
+The `deploy_info.json` file in the `vSwarm_deploy_metadata.tar.gz`  is used to identify the relative file paths for the Knative YAML manifests for deploying vSwarm functions. It also contains the path of YAML files needed as part of the pre-deployment commands to run certain vSwarm benchmarks, for example the `online-shop-database` which requires to be deployed before running `cartservice` benchmark.
+
+The `deploy_info.json` has the following schema:
+```console
+{
+    vswarm-function-name:
+        {
+            YamlLocation: /path/to/yaml
+            PredeploymentPath: [/path/to/predeployment-database/yaml]
+        }
+}
+```
+
+The `PredeploymentPath` is the path to the YAML file, which is applied via `kubectl apply -f`, before creating the service under `YamlLocation`. 
+
+## Deployment File Generation
+
+While the `deploy_info.json` file ships with the `vSwarm_deploy_metadata.tar.gz`, In order to regenerate the `deploy_info.json` run from the root of this repository:
+```console
+tar -xzvf workloads/container/vSwarm_deploy_metadata.tar.gz -C workloads/container
+cd workloads/container/
+python3 generate_deploy_info.py
+```
+
+## YAML Generation
+
+The `workloads/container/generate_all_yamls.py` is a wrapper script that calls the `generate-yamls.py` script for each vSwarm benchmark in the `vSwarm_deploy_metadata.tar.gz`. The `generate-yamls.py` Python script generates YAML files for vSwarm benchmarks with different workload parameters to create a variety of durations and memory consumption. 
+
+While the YAMLs are pre-generated inside the `vSwarm_deploy_metadata.tar.gz` tarball, to regenerate the YAMLs, run this from the root of this repository:
+```console
+tar -xzvf workloads/container/vSwarm_deploy_metadata.tar.gz -C workloads/container
+cd workloads/container
+python3 generate_all_yamls.py
+```

docs/loader.md

@@ -109,20 +109,42 @@ value previously collected.
 To account for difference in CPU performance set `ITERATIONS_MULTIPLIER=102` if using
 Cloudlab `xl170` or `d430` machines. (Date of measurement: 18-Oct-2022)
 
+## Executing vSwarm functions
+If you would like to use vSwarm benchmarks as profile functions to execute, first you need to generate a `mapper_output.json` using the `mapper` tool. Please refer to `mapper.md` docs for usage of the mapper tool to generate an output file. Once the `mapper_output.json` has been generated in the input trace directory, next run the following from the root of this repository:
+
+```console
+# install pre-requisites
+sudo apt update
+sudo apt -y install git-lfs pip xz-utils
+git lfs install
+git lfs fetch
+git lfs checkout
+```
+This retrieves the `vSwarm_deploy_metadata.tar.gz` from Git LFS. Then, untar this tarball by running the following command from the root of this repository:
+
+```bash
+$ tar -xzvf workloads/container/vSwarm_deploy_metadata.tar.gz -C workloads/container/
+```
+This untar the tarball into the `workloads/container/yamls` directory, which contains deployment information and deployment YAML files for all vSwarm benchmarks. If you would like to change some of these deployment files of regenerate them, refer to `generate_deploy_info_docs.md` for an outline of what these deployment files are, and how you can regenerate them.
+
 ## Single execution
 
 To run load generator use the following command:
 
 ```bash
 $ go run cmd/loader.go --config cmd/config_knative_trace.json
 ```
+To run load generator with vSwarm functions based on `mapper_output.json` run the following:
+
+```bash
+$ go run cmd/loader.go --config cmd/config_vswarm_trace.json
+```
+To direct the loader to execute vSwarm functions, set the `VSwarm` flag in the loader configuration `true`. For information on how to configure the workload for load generator, please refer to `docs/configuration.md`.
 
 Additionally, one can specify log verbosity argument as `--verbosity [info, debug, trace]`. The default value is `info`.
 
 To execute in a dry run mode without generating any load, set the `--dry-run` flag to `true`. This is useful for testing and validating configurations without executing actual requests.
 
-For to configure the workload for load generator, please refer to `docs/configuration.md`.
-
 There are a couple of constants that should not be exposed to the users. They can be examined and changed
 in `pkg/common/constants.go`.
 

go.mod

@@ -33,6 +33,7 @@ require (
 	github.com/campoy/embedmd v1.0.0 // indirect
 	github.com/davecgh/go-spew v1.1.1 // indirect
 	github.com/davidmz/go-pageant v1.0.2 // indirect
+	github.com/go-cmd/cmd v1.4.3 // indirect
 	github.com/go-fonts/liberation v0.3.3 // indirect
 	github.com/go-latex/latex v0.0.0-20240709081214-31cef3c7570e // indirect
 	github.com/go-logr/logr v1.4.2 // indirect

go.sum

@@ -18,6 +18,8 @@ github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davidmz/go-pageant v1.0.2 h1:bPblRCh5jGU+Uptpz6LgMZGD5hJoOt7otgT454WvHn0=
 github.com/davidmz/go-pageant v1.0.2/go.mod h1:P2EDDnMqIwG5Rrp05dTRITj9z2zpGcD9efWSkTNKLIE=
+github.com/go-cmd/cmd v1.4.3 h1:6y3G+3UqPerXvPcXvj+5QNPHT02BUw7p6PsqRxLNA7Y=
+github.com/go-cmd/cmd v1.4.3/go.mod h1:u3hxg/ry+D5kwh8WvUkHLAMe2zQCaXd00t35WfQaOFk=
 github.com/go-fonts/dejavu v0.3.4 h1:Qqyx9IOs5CQFxyWTdvddeWzrX0VNwUAvbmAzL0fpjbc=
 github.com/go-fonts/dejavu v0.3.4/go.mod h1:D1z0DglIz+lmpeNYMYlxW4r22IhcdOYnt+R3PShU/Kg=
 github.com/go-fonts/latin-modern v0.3.3 h1:g2xNgI8yzdNzIVm+qvbMryB6yGPe0pSMss8QT3QwlJ0=

pkg/common/trace_types.go

@@ -110,8 +110,9 @@ type Function struct {
 	CPURequestsMilli  int
 	MemoryRequestsMiB int
 	CPULimitsMilli    int
-
-	Specification *FunctionSpecification
+	YAMLPath          string
+	PredeploymentPath []string
+	Specification     *FunctionSpecification
 
 	// used only for dirigent workflows
 	WorkflowMetadata *WorkflowMetadata

pkg/config/parser_test.go

@@ -54,7 +54,7 @@ func TestConfigParser(t *testing.T) {
 		!strings.HasPrefix(config.OutputPathPrefix, "data/out/experiment") ||
 		config.IATDistribution != "equidistant" ||
 		config.CPULimit != "1vCPU" ||
-		config.ExperimentDuration != 5 ||
+		config.ExperimentDuration != 2 ||
 		config.WarmupDuration != 0 ||
 		config.IsPartiallyPanic != false ||
 		config.EnableZipkinTracing != false ||

pkg/config/test_config.json

@@ -11,7 +11,7 @@
   "OutputPathPrefix": "data/out/experiment",
   "IATDistribution": "equidistant",
   "CPULimit": "1vCPU",
-  "ExperimentDuration": 5,
+  "ExperimentDuration": 2,
   "WarmupDuration": 0,
 
   "IsPartiallyPanic": false,

pkg/config/test_vswarm_config.json

@@ -0,0 +1,32 @@
+{
+  "Seed": 42,
+
+  "Platform": "Knative",
+  "InvokeProtocol" : "grpc",
+  "YAMLSelector": "container",
+  "EndpointPort": 80,
+
+  "BusyLoopOnSandboxStartup": false,
+
+  "TracePath": "data/traces/example/",
+  "Granularity": "minute",
+  "OutputPathPrefix": "data/out/experiment",
+  "IATDistribution": "exponential",
+  "CPULimit": "1vCPU",
+  "ExperimentDuration": 2,
+  "WarmupDuration": 0,
+
+  "IsPartiallyPanic": false,
+  "EnableZipkinTracing": false,
+  "EnableMetricsScrapping": false,
+  "MetricScrapingPeriodSeconds": 15,
+  "AutoscalingMetric": "concurrency",
+
+  "GRPCConnectionTimeoutSeconds": 15,
+  "GRPCFunctionTimeoutSeconds": 900,
+  "DAGMode": false,
+  "EnableDAGDataset": true,
+  "Width": 2,
+  "Depth": 2,
+  "VSwarm": true
+}

pkg/driver/clients/grpc_client_test.go

@@ -25,19 +25,16 @@
 package clients
 
 import (
-	"context"
 	"fmt"
+	"os"
+	"testing"
+	"time"
+
 	"github.com/sirupsen/logrus"
 	"github.com/vhive-serverless/loader/pkg/common"
 	"github.com/vhive-serverless/loader/pkg/config"
 	"github.com/vhive-serverless/loader/pkg/workload/standard"
-	helloworld "github.com/vhive-serverless/vSwarm/utils/protobuf/helloworld"
-	"google.golang.org/grpc"
-	"google.golang.org/grpc/reflection"
-	"net"
-	"os"
-	"testing"
-	"time"
+	"github.com/vhive-serverless/loader/pkg/workload/vswarm"
 )
 
 func createFakeLoaderConfiguration() *config.LoaderConfiguration {
@@ -71,29 +68,6 @@ var testRuntimeSpecs = common.RuntimeSpecification{
 	Memory:  128,
 }
 
-type vSwarmServer struct {
-	helloworld.UnimplementedGreeterServer
-}
-
-func (s *vSwarmServer) SayHello(_ context.Context, req *helloworld.HelloRequest) (*helloworld.HelloReply, error) {
-	return &helloworld.HelloReply{
-		Message: "Reply message",
-	}, nil
-}
-
-func startVSwarmGRPCServer(serverAddress string, serverPort int) {
-	lis, err := net.Listen("tcp", fmt.Sprintf("%s:%d", serverAddress, serverPort))
-	if err != nil {
-		logrus.Fatalf("failed to listen: %v", err)
-	}
-
-	grpcServer := grpc.NewServer()
-
-	reflection.Register(grpcServer) // gRPC Server Reflection is used by gRPC CLI
-	helloworld.RegisterGreeterServer(grpcServer, &vSwarmServer{})
-	_ = grpcServer.Serve(lis)
-}
-
 func TestGRPCClientWithServerUnreachable(t *testing.T) {
 	cfg := createFakeLoaderConfiguration()
 	cfg.EnableZipkinTracing = true
@@ -161,7 +135,7 @@ func TestVSwarmClientWithServerReachable(t *testing.T) {
 	address, port := "localhost", 18081
 	testFunction.Endpoint = fmt.Sprintf("%s:%d", address, port)
 
-	go startVSwarmGRPCServer(address, port)
+	go vswarm.StartVSwarmGRPCServer(address, port)
 	time.Sleep(2 * time.Second)
 
 	cfgSwarm := createFakeVSwarmLoaderConfiguration()