// Copyright (C) NHR@FAU, University Erlangen-Nuremberg. // All rights reserved. // Use of this source code is governed by a MIT-style // license that can be found in the LICENSE file. package api import ( "bufio" "database/sql" "encoding/json" "errors" "fmt" "io" "net/http" "os" "path/filepath" "strconv" "strings" "sync" "time" "github.com/ClusterCockpit/cc-backend/internal/archiver" "github.com/ClusterCockpit/cc-backend/internal/auth" "github.com/ClusterCockpit/cc-backend/internal/config" "github.com/ClusterCockpit/cc-backend/internal/graph" "github.com/ClusterCockpit/cc-backend/internal/graph/model" "github.com/ClusterCockpit/cc-backend/internal/importer" "github.com/ClusterCockpit/cc-backend/internal/metricDataDispatcher" "github.com/ClusterCockpit/cc-backend/internal/repository" "github.com/ClusterCockpit/cc-backend/internal/util" "github.com/ClusterCockpit/cc-backend/pkg/archive" "github.com/ClusterCockpit/cc-backend/pkg/log" "github.com/ClusterCockpit/cc-backend/pkg/schema" "github.com/gorilla/mux" ) // @title ClusterCockpit REST API // @version 1.0.0 // @description API for batch job control. // @contact.name ClusterCockpit Project // @contact.url https://github.com/ClusterCockpit // @contact.email support@clustercockpit.org // @license.name MIT License // @license.url https://opensource.org/licenses/MIT // @host localhost:8080 // @basePath /api // @securityDefinitions.apikey ApiKeyAuth // @in header // @name X-Auth-Token type RestApi struct { JobRepository *repository.JobRepository Authentication *auth.Authentication MachineStateDir string RepositoryMutex sync.Mutex } func New() *RestApi { return &RestApi{ JobRepository: repository.GetJobRepository(), MachineStateDir: config.Keys.MachineStateDir, Authentication: auth.GetAuthInstance(), } } func (api *RestApi) MountApiRoutes(r *mux.Router) { r.StrictSlash(true) r.HandleFunc("/jobs/start_job/", api.startJob).Methods(http.MethodPost, http.MethodPut) r.HandleFunc("/jobs/stop_job/", api.stopJobByRequest).Methods(http.MethodPost, http.MethodPut) // r.HandleFunc("/jobs/import/", api.importJob).Methods(http.MethodPost, http.MethodPut) r.HandleFunc("/jobs/", api.getJobs).Methods(http.MethodGet) r.HandleFunc("/jobs/{id}", api.getJobById).Methods(http.MethodPost) r.HandleFunc("/jobs/{id}", api.getCompleteJobById).Methods(http.MethodGet) r.HandleFunc("/jobs/tag_job/{id}", api.tagJob).Methods(http.MethodPost, http.MethodPatch) r.HandleFunc("/jobs/edit_meta/{id}", api.editMeta).Methods(http.MethodPost, http.MethodPatch) r.HandleFunc("/jobs/metrics/{id}", api.getJobMetrics).Methods(http.MethodGet) r.HandleFunc("/jobs/delete_job/", api.deleteJobByRequest).Methods(http.MethodDelete) r.HandleFunc("/jobs/delete_job/{id}", api.deleteJobById).Methods(http.MethodDelete) r.HandleFunc("/jobs/delete_job_before/{ts}", api.deleteJobBefore).Methods(http.MethodDelete) r.HandleFunc("/clusters/", api.getClusters).Methods(http.MethodGet) if api.MachineStateDir != "" { r.HandleFunc("/machine_state/{cluster}/{host}", api.getMachineState).Methods(http.MethodGet) r.HandleFunc("/machine_state/{cluster}/{host}", api.putMachineState).Methods(http.MethodPut, http.MethodPost) } } func (api *RestApi) MountUserApiRoutes(r *mux.Router) { r.StrictSlash(true) r.HandleFunc("/jobs/", api.getJobs).Methods(http.MethodGet) r.HandleFunc("/jobs/{id}", api.getJobById).Methods(http.MethodPost) r.HandleFunc("/jobs/{id}", api.getCompleteJobById).Methods(http.MethodGet) r.HandleFunc("/jobs/metrics/{id}", api.getJobMetrics).Methods(http.MethodGet) } func (api *RestApi) MountConfigApiRoutes(r *mux.Router) { r.StrictSlash(true) if api.Authentication != nil { r.HandleFunc("/roles/", api.getRoles).Methods(http.MethodGet) r.HandleFunc("/users/", api.createUser).Methods(http.MethodPost, http.MethodPut) r.HandleFunc("/users/", api.getUsers).Methods(http.MethodGet) r.HandleFunc("/users/", api.deleteUser).Methods(http.MethodDelete) r.HandleFunc("/user/{id}", api.updateUser).Methods(http.MethodPost) r.HandleFunc("/notice/", api.editNotice).Methods(http.MethodPost) } } func (api *RestApi) MountFrontendApiRoutes(r *mux.Router) { r.StrictSlash(true) if api.Authentication != nil { r.HandleFunc("/jwt/", api.getJWT).Methods(http.MethodGet) r.HandleFunc("/configuration/", api.updateConfiguration).Methods(http.MethodPost) } } // DefaultApiResponse model type DefaultJobApiResponse struct { Message string `json:"msg"` } // StopJobApiRequest model type StopJobApiRequest struct { JobId *int64 `json:"jobId" example:"123000"` Cluster *string `json:"cluster" example:"fritz"` StartTime *int64 `json:"startTime" example:"1649723812"` State schema.JobState `json:"jobState" validate:"required" example:"completed"` StopTime int64 `json:"stopTime" validate:"required" example:"1649763839"` } // DeleteJobApiRequest model type DeleteJobApiRequest struct { JobId *int64 `json:"jobId" validate:"required" example:"123000"` // Cluster Job ID of job Cluster *string `json:"cluster" example:"fritz"` // Cluster of job StartTime *int64 `json:"startTime" example:"1649723812"` // Start Time of job as epoch } // GetJobsApiResponse model type GetJobsApiResponse struct { Jobs []*schema.JobMeta `json:"jobs"` // Array of jobs Items int `json:"items"` // Number of jobs returned Page int `json:"page"` // Page id returned } // GetClustersApiResponse model type GetClustersApiResponse struct { Clusters []*schema.Cluster `json:"clusters"` // Array of clusters } // ErrorResponse model type ErrorResponse struct { // Statustext of Errorcode Status string `json:"status"` Error string `json:"error"` // Error Message } // ApiTag model type ApiTag struct { // Tag Type Type string `json:"type" example:"Debug"` Name string `json:"name" example:"Testjob"` // Tag Name Scope string `json:"scope" example:"global"` // Tag Scope for Frontend Display } // ApiMeta model type EditMetaRequest struct { Key string `json:"key" example:"jobScript"` Value string `json:"value" example:"bash script"` } type TagJobApiRequest []*ApiTag type GetJobApiRequest []string type GetJobApiResponse struct { Meta *schema.Job Data []*JobMetricWithName } type GetCompleteJobApiResponse struct { Meta *schema.Job Data schema.JobData } type JobMetricWithName struct { Metric *schema.JobMetric `json:"metric"` Name string `json:"name"` Scope schema.MetricScope `json:"scope"` } type ApiReturnedUser struct { Username string `json:"username"` Name string `json:"name"` Roles []string `json:"roles"` Email string `json:"email"` Projects []string `json:"projects"` } func handleError(err error, statusCode int, rw http.ResponseWriter) { log.Warnf("REST ERROR : %s", err.Error()) rw.Header().Add("Content-Type", "application/json") rw.WriteHeader(statusCode) json.NewEncoder(rw).Encode(ErrorResponse{ Status: http.StatusText(statusCode), Error: err.Error(), }) } func decode(r io.Reader, val interface{}) error { dec := json.NewDecoder(r) dec.DisallowUnknownFields() return dec.Decode(val) } func securedCheck(r *http.Request) error { user := repository.GetUserFromContext(r.Context()) if user == nil { return fmt.Errorf("no user in context") } if user.AuthType == schema.AuthToken { // If nothing declared in config: deny all request to this endpoint if config.Keys.ApiAllowedIPs == nil || len(config.Keys.ApiAllowedIPs) == 0 { return fmt.Errorf("missing configuration key ApiAllowedIPs") } if config.Keys.ApiAllowedIPs[0] == "*" { return nil } // extract IP address IPAddress := r.Header.Get("X-Real-Ip") if IPAddress == "" { IPAddress = r.Header.Get("X-Forwarded-For") } if IPAddress == "" { IPAddress = r.RemoteAddr } if strings.Contains(IPAddress, ":") { IPAddress = strings.Split(IPAddress, ":")[0] } // check if IP is allowed if !util.Contains(config.Keys.ApiAllowedIPs, IPAddress) { return fmt.Errorf("unknown ip: %v", IPAddress) } } return nil } // getClusters godoc // @summary Lists all cluster configs // @tags Cluster query // @description Get a list of all cluster configs. Specific cluster can be requested using query parameter. // @produce json // @param cluster query string false "Job Cluster" // @success 200 {object} api.GetClustersApiResponse "Array of clusters" // @failure 400 {object} api.ErrorResponse "Bad Request" // @failure 401 {object} api.ErrorResponse "Unauthorized" // @failure 403 {object} api.ErrorResponse "Forbidden" // @failure 500 {object} api.ErrorResponse "Internal Server Error" // @security ApiKeyAuth // @router /clusters/ [get] func (api *RestApi) getClusters(rw http.ResponseWriter, r *http.Request) { if user := repository.GetUserFromContext(r.Context()); user != nil && !user.HasRole(schema.RoleApi) { handleError(fmt.Errorf("missing role: %v", schema.GetRoleString(schema.RoleApi)), http.StatusForbidden, rw) return } rw.Header().Add("Content-Type", "application/json") bw := bufio.NewWriter(rw) defer bw.Flush() var clusters []*schema.Cluster if r.URL.Query().Has("cluster") { name := r.URL.Query().Get("cluster") cluster := archive.GetCluster(name) if cluster == nil { handleError(fmt.Errorf("unknown cluster: %s", name), http.StatusBadRequest, rw) return } clusters = append(clusters, cluster) } else { clusters = archive.Clusters } payload := GetClustersApiResponse{ Clusters: clusters, } if err := json.NewEncoder(bw).Encode(payload); err != nil { handleError(err, http.StatusInternalServerError, rw) return } } // getJobs godoc // @summary Lists all jobs // @tags Job query // @description Get a list of all jobs. Filters can be applied using query parameters. // @description Number of results can be limited by page. Results are sorted by descending startTime. // @produce json // @param state query string false "Job State" Enums(running, completed, failed, cancelled, stopped, timeout) // @param cluster query string false "Job Cluster" // @param start-time query string false "Syntax: '$from-$to', as unix epoch timestamps in seconds" // @param items-per-page query int false "Items per page (Default: 25)" // @param page query int false "Page Number (Default: 1)" // @param with-metadata query bool false "Include metadata (e.g. jobScript) in response" // @success 200 {object} api.GetJobsApiResponse "Job array and page info" // @failure 400 {object} api.ErrorResponse "Bad Request" // @failure 401 {object} api.ErrorResponse "Unauthorized" // @failure 403 {object} api.ErrorResponse "Forbidden" // @failure 500 {object} api.ErrorResponse "Internal Server Error" // @security ApiKeyAuth // @router /jobs/ [get] func (api *RestApi) getJobs(rw http.ResponseWriter, r *http.Request) { withMetadata := false filter := &model.JobFilter{} page := &model.PageRequest{ItemsPerPage: 25, Page: 1} order := &model.OrderByInput{Field: "startTime", Type: "col", Order: model.SortDirectionEnumDesc} for key, vals := range r.URL.Query() { switch key { case "state": for _, s := range vals { state := schema.JobState(s) if !state.Valid() { handleError(fmt.Errorf("invalid query parameter value: state"), http.StatusBadRequest, rw) return } filter.State = append(filter.State, state) } case "cluster": filter.Cluster = &model.StringInput{Eq: &vals[0]} case "start-time": st := strings.Split(vals[0], "-") if len(st) != 2 { handleError(fmt.Errorf("invalid query parameter value: startTime"), http.StatusBadRequest, rw) return } from, err := strconv.ParseInt(st[0], 10, 64) if err != nil { handleError(err, http.StatusBadRequest, rw) return } to, err := strconv.ParseInt(st[1], 10, 64) if err != nil { handleError(err, http.StatusBadRequest, rw) return } ufrom, uto := time.Unix(from, 0), time.Unix(to, 0) filter.StartTime = &schema.TimeRange{From: &ufrom, To: &uto} case "page": x, err := strconv.Atoi(vals[0]) if err != nil { handleError(err, http.StatusBadRequest, rw) return } page.Page = x case "items-per-page": x, err := strconv.Atoi(vals[0]) if err != nil { handleError(err, http.StatusBadRequest, rw) return } page.ItemsPerPage = x case "with-metadata": withMetadata = true default: handleError(fmt.Errorf("invalid query parameter: %s", key), http.StatusBadRequest, rw) return } } jobs, err := api.JobRepository.QueryJobs(r.Context(), []*model.JobFilter{filter}, page, order) if err != nil { handleError(err, http.StatusInternalServerError, rw) return } results := make([]*schema.JobMeta, 0, len(jobs)) for _, job := range jobs { if withMetadata { if _, err = api.JobRepository.FetchMetadata(job); err != nil { handleError(err, http.StatusInternalServerError, rw) return } } res := &schema.JobMeta{ ID: &job.ID, BaseJob: job.BaseJob, StartTime: job.StartTime.Unix(), } res.Tags, err = api.JobRepository.GetTags(repository.GetUserFromContext(r.Context()), &job.ID) if err != nil { handleError(err, http.StatusInternalServerError, rw) return } if res.MonitoringStatus == schema.MonitoringStatusArchivingSuccessful { res.Statistics, err = archive.GetStatistics(job) if err != nil { handleError(err, http.StatusInternalServerError, rw) return } } results = append(results, res) } log.Debugf("/api/jobs: %d jobs returned", len(results)) rw.Header().Add("Content-Type", "application/json") bw := bufio.NewWriter(rw) defer bw.Flush() payload := GetJobsApiResponse{ Jobs: results, Items: page.ItemsPerPage, Page: page.Page, } if err := json.NewEncoder(bw).Encode(payload); err != nil { handleError(err, http.StatusInternalServerError, rw) return } } // getCompleteJobById godoc // @summary Get job meta and optional all metric data // @tags Job query // @description Job to get is specified by database ID // @description Returns full job resource information according to 'JobMeta' scheme and all metrics according to 'JobData'. // @produce json // @param id path int true "Database ID of Job" // @param all-metrics query bool false "Include all available metrics" // @success 200 {object} api.GetJobApiResponse "Job resource" // @failure 400 {object} api.ErrorResponse "Bad Request" // @failure 401 {object} api.ErrorResponse "Unauthorized" // @failure 403 {object} api.ErrorResponse "Forbidden" // @failure 404 {object} api.ErrorResponse "Resource not found" // @failure 422 {object} api.ErrorResponse "Unprocessable Entity: finding job failed: sql: no rows in result set" // @failure 500 {object} api.ErrorResponse "Internal Server Error" // @security ApiKeyAuth // @router /jobs/{id} [get] func (api *RestApi) getCompleteJobById(rw http.ResponseWriter, r *http.Request) { // Fetch job from db id, ok := mux.Vars(r)["id"] var job *schema.Job var err error if ok { id, e := strconv.ParseInt(id, 10, 64) if e != nil { handleError(fmt.Errorf("integer expected in path for id: %w", e), http.StatusBadRequest, rw) return } job, err = api.JobRepository.FindById(r.Context(), id) // Get Job from Repo by ID } else { handleError(fmt.Errorf("the parameter 'id' is required"), http.StatusBadRequest, rw) return } if err != nil { handleError(fmt.Errorf("finding job with db id %s failed: %w", id, err), http.StatusUnprocessableEntity, rw) return } job.Tags, err = api.JobRepository.GetTags(repository.GetUserFromContext(r.Context()), &job.ID) if err != nil { handleError(err, http.StatusInternalServerError, rw) return } if _, err = api.JobRepository.FetchMetadata(job); err != nil { handleError(err, http.StatusInternalServerError, rw) return } var scopes []schema.MetricScope if job.NumNodes == 1 { scopes = []schema.MetricScope{"core"} } else { scopes = []schema.MetricScope{"node"} } var data schema.JobData metricConfigs := archive.GetCluster(job.Cluster).MetricConfig resolution := 0 for _, mc := range metricConfigs { resolution = max(resolution, mc.Timestep) } if r.URL.Query().Get("all-metrics") == "true" { data, err = metricDataDispatcher.LoadData(job, nil, scopes, r.Context(), resolution) if err != nil { log.Warnf("REST: error while loading all-metrics job data for JobID %d on %s", job.JobID, job.Cluster) return } } log.Debugf("/api/job/%s: get job %d", id, job.JobID) rw.Header().Add("Content-Type", "application/json") bw := bufio.NewWriter(rw) defer bw.Flush() payload := GetCompleteJobApiResponse{ Meta: job, Data: data, } if err := json.NewEncoder(bw).Encode(payload); err != nil { handleError(err, http.StatusInternalServerError, rw) return } } // getJobById godoc // @summary Get job meta and configurable metric data // @tags Job query // @description Job to get is specified by database ID // @description Returns full job resource information according to 'JobMeta' scheme and all metrics according to 'JobData'. // @accept json // @produce json // @param id path int true "Database ID of Job" // @param request body api.GetJobApiRequest true "Array of metric names" // @success 200 {object} api.GetJobApiResponse "Job resource" // @failure 400 {object} api.ErrorResponse "Bad Request" // @failure 401 {object} api.ErrorResponse "Unauthorized" // @failure 403 {object} api.ErrorResponse "Forbidden" // @failure 404 {object} api.ErrorResponse "Resource not found" // @failure 422 {object} api.ErrorResponse "Unprocessable Entity: finding job failed: sql: no rows in result set" // @failure 500 {object} api.ErrorResponse "Internal Server Error" // @security ApiKeyAuth // @router /jobs/{id} [post] func (api *RestApi) getJobById(rw http.ResponseWriter, r *http.Request) { // Fetch job from db id, ok := mux.Vars(r)["id"] var job *schema.Job var err error if ok { id, e := strconv.ParseInt(id, 10, 64) if e != nil { handleError(fmt.Errorf("integer expected in path for id: %w", e), http.StatusBadRequest, rw) return } job, err = api.JobRepository.FindById(r.Context(), id) } else { handleError(errors.New("the parameter 'id' is required"), http.StatusBadRequest, rw) return } if err != nil { handleError(fmt.Errorf("finding job with db id %s failed: %w", id, err), http.StatusUnprocessableEntity, rw) return } job.Tags, err = api.JobRepository.GetTags(repository.GetUserFromContext(r.Context()), &job.ID) if err != nil { handleError(err, http.StatusInternalServerError, rw) return } if _, err = api.JobRepository.FetchMetadata(job); err != nil { handleError(err, http.StatusInternalServerError, rw) return } var metrics GetJobApiRequest if err = decode(r.Body, &metrics); err != nil { http.Error(rw, err.Error(), http.StatusBadRequest) return } var scopes []schema.MetricScope if job.NumNodes == 1 { scopes = []schema.MetricScope{"core"} } else { scopes = []schema.MetricScope{"node"} } metricConfigs := archive.GetCluster(job.Cluster).MetricConfig resolution := 0 for _, mc := range metricConfigs { resolution = max(resolution, mc.Timestep) } data, err := metricDataDispatcher.LoadData(job, metrics, scopes, r.Context(), resolution) if err != nil { log.Warnf("REST: error while loading job data for JobID %d on %s", job.JobID, job.Cluster) return } res := []*JobMetricWithName{} for name, md := range data { for scope, metric := range md { res = append(res, &JobMetricWithName{ Name: name, Scope: scope, Metric: metric, }) } } log.Debugf("/api/job/%s: get job %d", id, job.JobID) rw.Header().Add("Content-Type", "application/json") bw := bufio.NewWriter(rw) defer bw.Flush() payload := GetJobApiResponse{ Meta: job, Data: res, } if err := json.NewEncoder(bw).Encode(payload); err != nil { handleError(err, http.StatusInternalServerError, rw) return } } // editMeta godoc // @summary Edit meta-data json // @tags Job add and modify // @description Edit key value pairs in job metadata json // @description If a key already exists its content will be overwritten // @accept json // @produce json // @param id path int true "Job Database ID" // @param request body api.EditMetaRequest true "Kay value pair to add" // @success 200 {object} schema.Job "Updated job resource" // @failure 400 {object} api.ErrorResponse "Bad Request" // @failure 401 {object} api.ErrorResponse "Unauthorized" // @failure 404 {object} api.ErrorResponse "Job does not exist" // @failure 500 {object} api.ErrorResponse "Internal Server Error" // @security ApiKeyAuth // @router /jobs/edit_meta/{id} [post] func (api *RestApi) editMeta(rw http.ResponseWriter, r *http.Request) { id, err := strconv.ParseInt(mux.Vars(r)["id"], 10, 64) if err != nil { http.Error(rw, err.Error(), http.StatusBadRequest) return } job, err := api.JobRepository.FindById(r.Context(), id) if err != nil { http.Error(rw, err.Error(), http.StatusNotFound) return } var req EditMetaRequest if err := decode(r.Body, &req); err != nil { http.Error(rw, err.Error(), http.StatusBadRequest) return } if err := api.JobRepository.UpdateMetadata(job, req.Key, req.Value); err != nil { http.Error(rw, err.Error(), http.StatusInternalServerError) return } rw.Header().Add("Content-Type", "application/json") rw.WriteHeader(http.StatusOK) json.NewEncoder(rw).Encode(job) } // tagJob godoc // @summary Adds one or more tags to a job // @tags Job add and modify // @description Adds tag(s) to a job specified by DB ID. Name and Type of Tag(s) can be chosen freely. // @description Tag Scope for frontend visibility will default to "global" if none entered, other options: "admin" or specific username. // @description If tagged job is already finished: Tag will be written directly to respective archive files. // @accept json // @produce json // @param id path int true "Job Database ID" // @param request body api.TagJobApiRequest true "Array of tag-objects to add" // @success 200 {object} schema.Job "Updated job resource" // @failure 400 {object} api.ErrorResponse "Bad Request" // @failure 401 {object} api.ErrorResponse "Unauthorized" // @failure 404 {object} api.ErrorResponse "Job or tag does not exist" // @failure 500 {object} api.ErrorResponse "Internal Server Error" // @security ApiKeyAuth // @router /jobs/tag_job/{id} [post] func (api *RestApi) tagJob(rw http.ResponseWriter, r *http.Request) { id, err := strconv.ParseInt(mux.Vars(r)["id"], 10, 64) if err != nil { http.Error(rw, err.Error(), http.StatusBadRequest) return } job, err := api.JobRepository.FindById(r.Context(), id) if err != nil { http.Error(rw, err.Error(), http.StatusNotFound) return } job.Tags, err = api.JobRepository.GetTags(repository.GetUserFromContext(r.Context()), &job.ID) if err != nil { http.Error(rw, err.Error(), http.StatusInternalServerError) return } var req TagJobApiRequest if err := decode(r.Body, &req); err != nil { http.Error(rw, err.Error(), http.StatusBadRequest) return } for _, tag := range req { tagId, err := api.JobRepository.AddTagOrCreate(repository.GetUserFromContext(r.Context()), job.ID, tag.Type, tag.Name, tag.Scope) if err != nil { http.Error(rw, err.Error(), http.StatusInternalServerError) return } job.Tags = append(job.Tags, &schema.Tag{ ID: tagId, Type: tag.Type, Name: tag.Name, Scope: tag.Scope, }) } rw.Header().Add("Content-Type", "application/json") rw.WriteHeader(http.StatusOK) json.NewEncoder(rw).Encode(job) } // startJob godoc // @summary Adds a new job as "running" // @tags Job add and modify // @description Job specified in request body will be saved to database as "running" with new DB ID. // @description Job specifications follow the 'JobMeta' scheme, API will fail to execute if requirements are not met. // @accept json // @produce json // @param request body schema.JobMeta true "Job to add" // @success 201 {object} api.StartJobApiResponse "Job added successfully" // @failure 400 {object} api.ErrorResponse "Bad Request" // @failure 401 {object} api.ErrorResponse "Unauthorized" // @failure 403 {object} api.ErrorResponse "Forbidden" // @failure 422 {object} api.ErrorResponse "Unprocessable Entity: The combination of jobId, clusterId and startTime does already exist" // @failure 500 {object} api.ErrorResponse "Internal Server Error" // @security ApiKeyAuth // @router /jobs/start_job/ [post] func (api *RestApi) startJob(rw http.ResponseWriter, r *http.Request) { req := schema.JobMeta{BaseJob: schema.JobDefaults} if err := decode(r.Body, &req); err != nil { handleError(fmt.Errorf("parsing request body failed: %w", err), http.StatusBadRequest, rw) return } if req.State == "" { req.State = schema.JobStateRunning } if err := importer.SanityChecks(&req.BaseJob); err != nil { handleError(err, http.StatusBadRequest, rw) return } // aquire lock to avoid race condition between API calls var unlockOnce sync.Once api.RepositoryMutex.Lock() defer unlockOnce.Do(api.RepositoryMutex.Unlock) // Check if combination of (job_id, cluster_id, start_time) already exists: jobs, err := api.JobRepository.FindAll(&req.JobID, &req.Cluster, nil) if err != nil && err != sql.ErrNoRows { handleError(fmt.Errorf("checking for duplicate failed: %w", err), http.StatusInternalServerError, rw) return } else if err == nil { for _, job := range jobs { if (req.StartTime - job.StartTimeUnix) < 86400 { handleError(fmt.Errorf("a job with that jobId, cluster and startTime already exists: dbid: %d, jobid: %d", job.ID, job.JobID), http.StatusUnprocessableEntity, rw) return } } } id, err := api.JobRepository.Start(&req) if err != nil { handleError(fmt.Errorf("insert into database failed: %w", err), http.StatusInternalServerError, rw) return } // unlock here, adding Tags can be async unlockOnce.Do(api.RepositoryMutex.Unlock) for _, tag := range req.Tags { if _, err := api.JobRepository.AddTagOrCreate(repository.GetUserFromContext(r.Context()), id, tag.Type, tag.Name, tag.Scope); err != nil { http.Error(rw, err.Error(), http.StatusInternalServerError) handleError(fmt.Errorf("adding tag to new job %d failed: %w", id, err), http.StatusInternalServerError, rw) return } } log.Printf("new job (id: %d): cluster=%s, jobId=%d, user=%s, startTime=%d", id, req.Cluster, req.JobID, req.User, req.StartTime) rw.Header().Add("Content-Type", "application/json") rw.WriteHeader(http.StatusCreated) json.NewEncoder(rw).Encode(DefaultJobApiResponse{ Message: "success", }) } // stopJobByRequest godoc // @summary Marks job as completed and triggers archiving // @tags Job add and modify // @description Job to stop is specified by request body. All fields are required in this case. // @description Returns full job resource information according to 'JobMeta' scheme. // @produce json // @param request body api.StopJobApiRequest true "All fields required" // @success 200 {object} schema.JobMeta "Success message" // @failure 400 {object} api.ErrorResponse "Bad Request" // @failure 401 {object} api.ErrorResponse "Unauthorized" // @failure 403 {object} api.ErrorResponse "Forbidden" // @failure 404 {object} api.ErrorResponse "Resource not found" // @failure 422 {object} api.ErrorResponse "Unprocessable Entity: finding job failed: sql: no rows in result set" // @failure 500 {object} api.ErrorResponse "Internal Server Error" // @security ApiKeyAuth // @router /jobs/stop_job/ [post] func (api *RestApi) stopJobByRequest(rw http.ResponseWriter, r *http.Request) { // Parse request body req := StopJobApiRequest{} if err := decode(r.Body, &req); err != nil { handleError(fmt.Errorf("parsing request body failed: %w", err), http.StatusBadRequest, rw) return } // Fetch job (that will be stopped) from db var job *schema.Job var err error if req.JobId == nil { handleError(errors.New("the field 'jobId' is required"), http.StatusBadRequest, rw) return } // log.Printf("loading db job for stopJobByRequest... : stopJobApiRequest=%v", req) job, err = api.JobRepository.Find(req.JobId, req.Cluster, req.StartTime) if err != nil { handleError(fmt.Errorf("finding job failed: %w", err), http.StatusUnprocessableEntity, rw) return } api.checkAndHandleStopJob(rw, job, req) } // deleteJobById godoc // @summary Remove a job from the sql database // @tags Job remove // @description Job to remove is specified by database ID. This will not remove the job from the job archive. // @produce json // @param id path int true "Database ID of Job" // @success 200 {object} api.DeleteJobApiResponse "Success message" // @failure 400 {object} api.ErrorResponse "Bad Request" // @failure 401 {object} api.ErrorResponse "Unauthorized" // @failure 403 {object} api.ErrorResponse "Forbidden" // @failure 404 {object} api.ErrorResponse "Resource not found" // @failure 422 {object} api.ErrorResponse "Unprocessable Entity: finding job failed: sql: no rows in result set" // @failure 500 {object} api.ErrorResponse "Internal Server Error" // @security ApiKeyAuth // @router /jobs/delete_job/{id} [delete] func (api *RestApi) deleteJobById(rw http.ResponseWriter, r *http.Request) { // Fetch job (that will be stopped) from db id, ok := mux.Vars(r)["id"] var err error if ok { id, e := strconv.ParseInt(id, 10, 64) if e != nil { handleError(fmt.Errorf("integer expected in path for id: %w", e), http.StatusBadRequest, rw) return } err = api.JobRepository.DeleteJobById(id) } else { handleError(errors.New("the parameter 'id' is required"), http.StatusBadRequest, rw) return } if err != nil { handleError(fmt.Errorf("deleting job failed: %w", err), http.StatusUnprocessableEntity, rw) return } rw.Header().Add("Content-Type", "application/json") rw.WriteHeader(http.StatusOK) json.NewEncoder(rw).Encode(DefaultJobApiResponse{ Message: fmt.Sprintf("Successfully deleted job %s", id), }) } // deleteJobByRequest godoc // @summary Remove a job from the sql database // @tags Job remove // @description Job to delete is specified by request body. All fields are required in this case. // @accept json // @produce json // @param request body api.DeleteJobApiRequest true "All fields required" // @success 200 {object} api.DeleteJobApiResponse "Success message" // @failure 400 {object} api.ErrorResponse "Bad Request" // @failure 401 {object} api.ErrorResponse "Unauthorized" // @failure 403 {object} api.ErrorResponse "Forbidden" // @failure 404 {object} api.ErrorResponse "Resource not found" // @failure 422 {object} api.ErrorResponse "Unprocessable Entity: finding job failed: sql: no rows in result set" // @failure 500 {object} api.ErrorResponse "Internal Server Error" // @security ApiKeyAuth // @router /jobs/delete_job/ [delete] func (api *RestApi) deleteJobByRequest(rw http.ResponseWriter, r *http.Request) { // Parse request body req := DeleteJobApiRequest{} if err := decode(r.Body, &req); err != nil { handleError(fmt.Errorf("parsing request body failed: %w", err), http.StatusBadRequest, rw) return } // Fetch job (that will be deleted) from db var job *schema.Job var err error if req.JobId == nil { handleError(errors.New("the field 'jobId' is required"), http.StatusBadRequest, rw) return } job, err = api.JobRepository.Find(req.JobId, req.Cluster, req.StartTime) if err != nil { handleError(fmt.Errorf("finding job failed: %w", err), http.StatusUnprocessableEntity, rw) return } err = api.JobRepository.DeleteJobById(job.ID) if err != nil { handleError(fmt.Errorf("deleting job failed: %w", err), http.StatusUnprocessableEntity, rw) return } rw.Header().Add("Content-Type", "application/json") rw.WriteHeader(http.StatusOK) json.NewEncoder(rw).Encode(DefaultJobApiResponse{ Message: fmt.Sprintf("Successfully deleted job %d", job.ID), }) } // deleteJobBefore godoc // @summary Remove a job from the sql database // @tags Job remove // @description Remove all jobs with start time before timestamp. The jobs will not be removed from the job archive. // @produce json // @param ts path int true "Unix epoch timestamp" // @success 200 {object} api.DeleteJobApiResponse "Success message" // @failure 400 {object} api.ErrorResponse "Bad Request" // @failure 401 {object} api.ErrorResponse "Unauthorized" // @failure 403 {object} api.ErrorResponse "Forbidden" // @failure 404 {object} api.ErrorResponse "Resource not found" // @failure 422 {object} api.ErrorResponse "Unprocessable Entity: finding job failed: sql: no rows in result set" // @failure 500 {object} api.ErrorResponse "Internal Server Error" // @security ApiKeyAuth // @router /jobs/delete_job_before/{ts} [delete] func (api *RestApi) deleteJobBefore(rw http.ResponseWriter, r *http.Request) { var cnt int // Fetch job (that will be stopped) from db id, ok := mux.Vars(r)["ts"] var err error if ok { ts, e := strconv.ParseInt(id, 10, 64) if e != nil { handleError(fmt.Errorf("integer expected in path for ts: %w", e), http.StatusBadRequest, rw) return } cnt, err = api.JobRepository.DeleteJobsBefore(ts) } else { handleError(errors.New("the parameter 'ts' is required"), http.StatusBadRequest, rw) return } if err != nil { handleError(fmt.Errorf("deleting jobs failed: %w", err), http.StatusUnprocessableEntity, rw) return } rw.Header().Add("Content-Type", "application/json") rw.WriteHeader(http.StatusOK) json.NewEncoder(rw).Encode(DefaultJobApiResponse{ Message: fmt.Sprintf("Successfully deleted %d jobs", cnt), }) } func (api *RestApi) checkAndHandleStopJob(rw http.ResponseWriter, job *schema.Job, req StopJobApiRequest) { // Sanity checks if job == nil || job.StartTime.Unix() >= req.StopTime || job.State != schema.JobStateRunning { handleError(fmt.Errorf("jobId %d (id %d) on %s : stopTime %d must be larger than startTime %d and only running jobs can be stopped (state is: %s)", job.JobID, job.ID, job.Cluster, req.StopTime, job.StartTime.Unix(), job.State), http.StatusBadRequest, rw) return } if req.State != "" && !req.State.Valid() { handleError(fmt.Errorf("jobId %d (id %d) on %s : invalid requested job state: %#v", job.JobID, job.ID, job.Cluster, req.State), http.StatusBadRequest, rw) return } else if req.State == "" { req.State = schema.JobStateCompleted } // Mark job as stopped in the database (update state and duration) job.Duration = int32(req.StopTime - job.StartTime.Unix()) job.State = req.State if err := api.JobRepository.Stop(job.ID, job.Duration, job.State, job.MonitoringStatus); err != nil { handleError(fmt.Errorf("jobId %d (id %d) on %s : marking job as '%s' (duration: %d) in DB failed: %w", job.JobID, job.ID, job.Cluster, job.State, job.Duration, err), http.StatusInternalServerError, rw) return } log.Printf("archiving job... (dbid: %d): cluster=%s, jobId=%d, user=%s, startTime=%s, duration=%d, state=%s", job.ID, job.Cluster, job.JobID, job.User, job.StartTime, job.Duration, job.State) // Send a response (with status OK). This means that erros that happen from here on forward // can *NOT* be communicated to the client. If reading from a MetricDataRepository or // writing to the filesystem fails, the client will not know. rw.Header().Add("Content-Type", "application/json") rw.WriteHeader(http.StatusOK) json.NewEncoder(rw).Encode(job) // Monitoring is disabled... if job.MonitoringStatus == schema.MonitoringStatusDisabled { return } // Trigger async archiving archiver.TriggerArchiving(job) } func (api *RestApi) getJobMetrics(rw http.ResponseWriter, r *http.Request) { id := mux.Vars(r)["id"] metrics := r.URL.Query()["metric"] var scopes []schema.MetricScope for _, scope := range r.URL.Query()["scope"] { var s schema.MetricScope if err := s.UnmarshalGQL(scope); err != nil { http.Error(rw, err.Error(), http.StatusBadRequest) return } scopes = append(scopes, s) } rw.Header().Add("Content-Type", "application/json") rw.WriteHeader(http.StatusOK) type Respone struct { Data *struct { JobMetrics []*model.JobMetricWithName `json:"jobMetrics"` } `json:"data"` Error *struct { Message string `json:"message"` } `json:"error"` } resolver := graph.GetResolverInstance() data, err := resolver.Query().JobMetrics(r.Context(), id, metrics, scopes, nil) if err != nil { json.NewEncoder(rw).Encode(Respone{ Error: &struct { Message string "json:\"message\"" }{Message: err.Error()}, }) return } json.NewEncoder(rw).Encode(Respone{ Data: &struct { JobMetrics []*model.JobMetricWithName "json:\"jobMetrics\"" }{JobMetrics: data}, }) } // createUser godoc // @summary Adds a new user // @tags User // @description User specified in form data will be saved to database. // @description Only accessible from IPs registered with apiAllowedIPs configuration option. // @accept mpfd // @produce plain // @param username formData string true "Unique user ID" // @param password formData string true "User password" // @param role formData string true "User role" Enums(admin, support, manager, user, api) // @param project formData string false "Managed project, required for new manager role user" // @param name formData string false "Users name" // @param email formData string false "Users email" // @success 200 {string} string "Success Response" // @failure 400 {string} string "Bad Request" // @failure 401 {string} string "Unauthorized" // @failure 403 {string} string "Forbidden" // @failure 422 {string} string "Unprocessable Entity: creating user failed" // @failure 500 {string} string "Internal Server Error" // @security ApiKeyAuth // @router /users/ [post] func (api *RestApi) createUser(rw http.ResponseWriter, r *http.Request) { err := securedCheck(r) if err != nil { http.Error(rw, err.Error(), http.StatusForbidden) return } rw.Header().Set("Content-Type", "text/plain") me := repository.GetUserFromContext(r.Context()) if !me.HasRole(schema.RoleAdmin) { http.Error(rw, "Only admins are allowed to create new users", http.StatusForbidden) return } username, password, role, name, email, project := r.FormValue("username"), r.FormValue("password"), r.FormValue("role"), r.FormValue("name"), r.FormValue("email"), r.FormValue("project") if len(password) == 0 && role != schema.GetRoleString(schema.RoleApi) { http.Error(rw, "Only API users are allowed to have a blank password (login will be impossible)", http.StatusBadRequest) return } if len(project) != 0 && role != schema.GetRoleString(schema.RoleManager) { http.Error(rw, "only managers require a project (can be changed later)", http.StatusBadRequest) return } else if len(project) == 0 && role == schema.GetRoleString(schema.RoleManager) { http.Error(rw, "managers require a project to manage (can be changed later)", http.StatusBadRequest) return } if err := repository.GetUserRepository().AddUser(&schema.User{ Username: username, Name: name, Password: password, Email: email, Projects: []string{project}, Roles: []string{role}, }); err != nil { http.Error(rw, err.Error(), http.StatusUnprocessableEntity) return } fmt.Fprintf(rw, "User %v successfully created!\n", username) } // deleteUser godoc // @summary Deletes a user // @tags User // @description User defined by username in form data will be deleted from database. // @description Only accessible from IPs registered with apiAllowedIPs configuration option. // @accept mpfd // @produce plain // @param username formData string true "User ID to delete" // @success 200 "User deleted successfully" // @failure 400 {string} string "Bad Request" // @failure 401 {string} string "Unauthorized" // @failure 403 {string} string "Forbidden" // @failure 422 {string} string "Unprocessable Entity: deleting user failed" // @failure 500 {string} string "Internal Server Error" // @security ApiKeyAuth // @router /users/ [delete] func (api *RestApi) deleteUser(rw http.ResponseWriter, r *http.Request) { err := securedCheck(r) if err != nil { http.Error(rw, err.Error(), http.StatusForbidden) return } if user := repository.GetUserFromContext(r.Context()); !user.HasRole(schema.RoleAdmin) { http.Error(rw, "Only admins are allowed to delete a user", http.StatusForbidden) return } username := r.FormValue("username") if err := repository.GetUserRepository().DelUser(username); err != nil { http.Error(rw, err.Error(), http.StatusUnprocessableEntity) return } rw.WriteHeader(http.StatusOK) } // getUsers godoc // @summary Returns a list of users // @tags User // @description Returns a JSON-encoded list of users. // @description Required query-parameter defines if all users or only users with additional special roles are returned. // @description Only accessible from IPs registered with apiAllowedIPs configuration option. // @produce json // @param not-just-user query bool true "If returned list should contain all users or only users with additional special roles" // @success 200 {array} api.ApiReturnedUser "List of users returned successfully" // @failure 400 {string} string "Bad Request" // @failure 401 {string} string "Unauthorized" // @failure 403 {string} string "Forbidden" // @failure 500 {string} string "Internal Server Error" // @security ApiKeyAuth // @router /users/ [get] func (api *RestApi) getUsers(rw http.ResponseWriter, r *http.Request) { err := securedCheck(r) if err != nil { http.Error(rw, err.Error(), http.StatusForbidden) return } if user := repository.GetUserFromContext(r.Context()); !user.HasRole(schema.RoleAdmin) { http.Error(rw, "Only admins are allowed to fetch a list of users", http.StatusForbidden) return } users, err := repository.GetUserRepository().ListUsers(r.URL.Query().Get("not-just-user") == "true") if err != nil { http.Error(rw, err.Error(), http.StatusInternalServerError) return } json.NewEncoder(rw).Encode(users) } // updateUser godoc // @summary Updates an existing user // @tags User // @description Modifies user defined by username (id) in one of four possible ways. // @description If more than one formValue is set then only the highest priority field is used. // @description Only accessible from IPs registered with apiAllowedIPs configuration option. // @accept mpfd // @produce plain // @param id path string true "Database ID of User" // @param add-role formData string false "Priority 1: Role to add" Enums(admin, support, manager, user, api) // @param remove-role formData string false "Priority 2: Role to remove" Enums(admin, support, manager, user, api) // @param add-project formData string false "Priority 3: Project to add" // @param remove-project formData string false "Priority 4: Project to remove" // @success 200 {string} string "Success Response Message" // @failure 400 {string} string "Bad Request" // @failure 401 {string} string "Unauthorized" // @failure 403 {string} string "Forbidden" // @failure 422 {string} string "Unprocessable Entity: The user could not be updated" // @failure 500 {string} string "Internal Server Error" // @security ApiKeyAuth // @router /user/{id} [post] func (api *RestApi) updateUser(rw http.ResponseWriter, r *http.Request) { err := securedCheck(r) if err != nil { http.Error(rw, err.Error(), http.StatusForbidden) return } if user := repository.GetUserFromContext(r.Context()); !user.HasRole(schema.RoleAdmin) { http.Error(rw, "Only admins are allowed to update a user", http.StatusForbidden) return } // Get Values newrole := r.FormValue("add-role") delrole := r.FormValue("remove-role") newproj := r.FormValue("add-project") delproj := r.FormValue("remove-project") // TODO: Handle anything but roles... if newrole != "" { if err := repository.GetUserRepository().AddRole(r.Context(), mux.Vars(r)["id"], newrole); err != nil { http.Error(rw, err.Error(), http.StatusUnprocessableEntity) return } rw.Write([]byte("Add Role Success")) } else if delrole != "" { if err := repository.GetUserRepository().RemoveRole(r.Context(), mux.Vars(r)["id"], delrole); err != nil { http.Error(rw, err.Error(), http.StatusUnprocessableEntity) return } rw.Write([]byte("Remove Role Success")) } else if newproj != "" { if err := repository.GetUserRepository().AddProject(r.Context(), mux.Vars(r)["id"], newproj); err != nil { http.Error(rw, err.Error(), http.StatusUnprocessableEntity) return } rw.Write([]byte("Add Project Success")) } else if delproj != "" { if err := repository.GetUserRepository().RemoveProject(r.Context(), mux.Vars(r)["id"], delproj); err != nil { http.Error(rw, err.Error(), http.StatusUnprocessableEntity) return } rw.Write([]byte("Remove Project Success")) } else { http.Error(rw, "Not Add or Del [role|project]?", http.StatusInternalServerError) } } // editNotice godoc // @summary Updates or empties the notice box content // @tags User // @description Modifies the content of notice.txt, shown as notice box on the homepage. // @description If more than one formValue is set then only the highest priority field is used. // @description Only accessible from IPs registered with apiAllowedIPs configuration option. // @accept mpfd // @produce plain // @param new-content formData string false "Priority 1: New content to display" // @success 200 {string} string "Success Response Message" // @failure 400 {string} string "Bad Request" // @failure 401 {string} string "Unauthorized" // @failure 403 {string} string "Forbidden" // @failure 422 {string} string "Unprocessable Entity: The user could not be updated" // @failure 500 {string} string "Internal Server Error" // @security ApiKeyAuth // @router /notice/ [post] func (api *RestApi) editNotice(rw http.ResponseWriter, r *http.Request) { err := securedCheck(r) if err != nil { http.Error(rw, err.Error(), http.StatusForbidden) return } if user := repository.GetUserFromContext(r.Context()); !user.HasRole(schema.RoleAdmin) { http.Error(rw, "Only admins are allowed to update the notice.txt file", http.StatusForbidden) return } // Get Value newContent := r.FormValue("new-content") // Check FIle noticeExists := util.CheckFileExists("./var/notice.txt") if !noticeExists { ntxt, err := os.Create("./var/notice.txt") if err != nil { log.Errorf("Creating ./var/notice.txt failed: %s", err.Error()) http.Error(rw, err.Error(), http.StatusUnprocessableEntity) return } ntxt.Close() } if newContent != "" { if err := os.WriteFile("./var/notice.txt", []byte(newContent), 0o666); err != nil { log.Errorf("Writing to ./var/notice.txt failed: %s", err.Error()) http.Error(rw, err.Error(), http.StatusUnprocessableEntity) return } else { rw.Write([]byte("Update Notice Content Success")) } } else { if err := os.WriteFile("./var/notice.txt", []byte(""), 0o666); err != nil { log.Errorf("Writing to ./var/notice.txt failed: %s", err.Error()) http.Error(rw, err.Error(), http.StatusUnprocessableEntity) return } else { rw.Write([]byte("Empty Notice Content Success")) } } } func (api *RestApi) getJWT(rw http.ResponseWriter, r *http.Request) { err := securedCheck(r) if err != nil { http.Error(rw, err.Error(), http.StatusForbidden) return } rw.Header().Set("Content-Type", "text/plain") username := r.FormValue("username") me := repository.GetUserFromContext(r.Context()) if !me.HasRole(schema.RoleAdmin) { if username != me.Username { http.Error(rw, "Only admins are allowed to sign JWTs not for themselves", http.StatusForbidden) return } } user, err := repository.GetUserRepository().GetUser(username) if err != nil { http.Error(rw, err.Error(), http.StatusUnprocessableEntity) return } jwt, err := api.Authentication.JwtAuth.ProvideJWT(user) if err != nil { http.Error(rw, err.Error(), http.StatusUnprocessableEntity) return } rw.WriteHeader(http.StatusOK) rw.Write([]byte(jwt)) } func (api *RestApi) getRoles(rw http.ResponseWriter, r *http.Request) { err := securedCheck(r) if err != nil { http.Error(rw, err.Error(), http.StatusForbidden) return } user := repository.GetUserFromContext(r.Context()) if !user.HasRole(schema.RoleAdmin) { http.Error(rw, "only admins are allowed to fetch a list of roles", http.StatusForbidden) return } roles, err := schema.GetValidRoles(user) if err != nil { http.Error(rw, err.Error(), http.StatusInternalServerError) return } json.NewEncoder(rw).Encode(roles) } func (api *RestApi) updateConfiguration(rw http.ResponseWriter, r *http.Request) { rw.Header().Set("Content-Type", "text/plain") key, value := r.FormValue("key"), r.FormValue("value") fmt.Printf("REST > KEY: %#v\nVALUE: %#v\n", key, value) if err := repository.GetUserCfgRepo().UpdateConfig(key, value, repository.GetUserFromContext(r.Context())); err != nil { http.Error(rw, err.Error(), http.StatusUnprocessableEntity) return } rw.Write([]byte("success")) } func (api *RestApi) putMachineState(rw http.ResponseWriter, r *http.Request) { if api.MachineStateDir == "" { http.Error(rw, "REST > machine state not enabled", http.StatusNotFound) return } vars := mux.Vars(r) cluster := vars["cluster"] host := vars["host"] dir := filepath.Join(api.MachineStateDir, cluster) if err := os.MkdirAll(dir, 0755); err != nil { http.Error(rw, err.Error(), http.StatusInternalServerError) return } filename := filepath.Join(dir, fmt.Sprintf("%s.json", host)) f, err := os.Create(filename) if err != nil { http.Error(rw, err.Error(), http.StatusInternalServerError) return } defer f.Close() if _, err := io.Copy(f, r.Body); err != nil { http.Error(rw, err.Error(), http.StatusInternalServerError) return } rw.WriteHeader(http.StatusCreated) } func (api *RestApi) getMachineState(rw http.ResponseWriter, r *http.Request) { if api.MachineStateDir == "" { http.Error(rw, "REST > machine state not enabled", http.StatusNotFound) return } vars := mux.Vars(r) filename := filepath.Join(api.MachineStateDir, vars["cluster"], fmt.Sprintf("%s.json", vars["host"])) // Sets the content-type and 'Last-Modified' Header and so on automatically http.ServeFile(rw, r, filename) }