// GetCluster returns a given cluster definition func GetCluster(w rest.ResponseWriter, r *rest.Request) { dbConn, err := util.GetConnection(CLUSTERADMIN_DB) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), 400) return } defer dbConn.Close() err = secimpl.Authorize(dbConn, r.PathParam("Token"), "perm-read") if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusUnauthorized) return } ID := r.PathParam("ID") results, err := admindb.GetCluster(dbConn, ID) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) } cluster := types.Cluster{} cluster.ID = results.ID cluster.ProjectID = results.ProjectID cluster.Name = results.Name cluster.ClusterType = results.ClusterType cluster.Status = results.Status cluster.CreateDate = results.CreateDate cluster.Containers = results.Containers //logit.Info.Println("GetCluser:db call results=" + results.ID) w.WriteJson(&cluster) }
// StartCluster starts all nodes in a cluster func StartCluster(w rest.ResponseWriter, r *rest.Request) { dbConn, err := util.GetConnection(CLUSTERADMIN_DB) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), 400) return } defer dbConn.Close() err = secimpl.Authorize(dbConn, r.PathParam("Token"), "perm-cluster") if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusUnauthorized) return } ID := r.PathParam("ID") if ID == "" { logit.Error.Println("StartCluster: error cluster ID required") rest.Error(w, "cluster ID required", http.StatusBadRequest) return } cluster, err := admindb.GetCluster(dbConn, ID) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) } //start docker containers containers, err := admindb.GetAllContainersForCluster(dbConn, cluster.ID) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) } i := 0 i = 0 var response swarmapi.DockerStartResponse for i = range containers { req := &swarmapi.DockerStartRequest{} req.ContainerName = containers[i].Name logit.Info.Println("will attempt to start container " + req.ContainerName) response, err = swarmapi.DockerStart(req) if err != nil { logit.Error.Println("StartCluster: error when trying to start container" + err.Error()) } logit.Info.Println("StartCluster: started " + response.Output) i++ } status := types.SimpleStatus{} status.Status = "OK" w.WriteHeader(http.StatusOK) w.WriteJson(&status) }
// TODO func ConfigureCluster(w rest.ResponseWriter, r *rest.Request) { dbConn, err := util.GetConnection(CLUSTERADMIN_DB) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), 400) return } defer dbConn.Close() err = secimpl.Authorize(dbConn, r.PathParam("Token"), "perm-cluster") if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusUnauthorized) return } ID := r.PathParam("ID") cluster, err := admindb.GetCluster(dbConn, ID) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } err = configureCluster("SM", dbConn, cluster, false) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } w.WriteHeader(http.StatusOK) status := types.SimpleStatus{} status.Status = "OK" w.WriteJson(&status) }
// AdminFailover causes a cluster failorver to be performed for a given cluster func AdminFailover(w rest.ResponseWriter, r *rest.Request) { dbConn, err := util.GetConnection(CLUSTERADMIN_DB) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), 400) return } defer dbConn.Close() err = secimpl.Authorize(dbConn, r.PathParam("Token"), "perm-cluster") if err != nil { logit.Error.Println("authorize error " + err.Error()) rest.Error(w, err.Error(), http.StatusUnauthorized) return } ID := r.PathParam("ID") if ID == "" { logit.Error.Println("node ID required error") rest.Error(w, "node ID required", http.StatusBadRequest) return } //dbNode is the standby node we are going to fail over and //make the new master in the cluster var dbNode types.Container dbNode, err = admindb.GetContainer(dbConn, ID) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } cluster, err := admindb.GetCluster(dbConn, dbNode.ClusterID) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } var failoverResp cpmcontainerapi.FailoverResponse failoverResp, err = cpmcontainerapi.FailoverClient(dbNode.Name) if err != nil { logit.Error.Println("fail-over error " + err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } logit.Info.Println("AdminFailover: fail-over output " + failoverResp.Output) //update the old master to standalone role oldMaster := types.Container{} oldMaster, err = admindb.GetContainerMaster(dbConn, dbNode.ClusterID) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } oldMaster.Role = "standalone" oldMaster.ClusterID = "-1" err = admindb.UpdateContainer(dbConn, oldMaster) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } //update the failover node to master role dbNode.Role = "master" err = admindb.UpdateContainer(dbConn, dbNode) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } //stop pg on the old master //params.IPAddress1 = oldMaster.IPAddress var stopPGResp cpmcontainerapi.StopPGResponse stopPGResp, err = cpmcontainerapi.StopPGClient(oldMaster.Name) if err != nil { logit.Error.Println(err.Error() + stopPGResp.Output) rest.Error(w, err.Error(), http.StatusBadRequest) return } err = configureCluster("SM", dbConn, cluster, false) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } w.WriteHeader(http.StatusOK) status := types.SimpleStatus{} status.Status = "OK" w.WriteJson(&status) return }
// DeleteCluster deletes an existing cluster definition func DeleteCluster(w rest.ResponseWriter, r *rest.Request) { dbConn, err := util.GetConnection(CLUSTERADMIN_DB) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), 400) return } defer dbConn.Close() err = secimpl.Authorize(dbConn, r.PathParam("Token"), "perm-cluster") if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusUnauthorized) return } ID := r.PathParam("ID") if ID == "" { logit.Error.Println("cluster ID required") rest.Error(w, "cluster ID required", http.StatusBadRequest) return } cluster, err := admindb.GetCluster(dbConn, ID) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) } //delete docker containers containers, err := admindb.GetAllContainersForCluster(dbConn, cluster.ID) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) } var infoResponse swarmapi.DockerInfoResponse infoResponse, err = swarmapi.DockerInfo() if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } i := 0 servers := make([]types.Server, len(infoResponse.Output)) for i = range infoResponse.Output { servers[i].ID = infoResponse.Output[i] servers[i].Name = infoResponse.Output[i] servers[i].IPAddress = infoResponse.Output[i] i++ } var pgdatapath types.Setting pgdatapath, err = admindb.GetSetting(dbConn, "PG-DATA-PATH") if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } i = 0 //server := types.Server{} for i = range containers { //logit.Info.Println("DeleteCluster: got server IP " + server.IPAddress) //it is possible that someone can remove a container //outside of us, so we let it pass that we can't remove //it //err = removeContainer(server.IPAddress, containers[i].Name) dremreq := &swarmapi.DockerRemoveRequest{} dremreq.ContainerName = containers[i].Name //logit.Info.Println("will attempt to delete container " + dremreq.ContainerName) _, err = swarmapi.DockerRemove(dremreq) if err != nil { logit.Error.Println("error when trying to remove container" + err.Error()) } //send all the servers a deletevolume command ddreq := &cpmserverapi.DiskDeleteRequest{} ddreq.Path = pgdatapath.Value + "/" + containers[i].Name for _, each := range servers { _, err = cpmserverapi.DiskDeleteClient(each.Name, ddreq) if err != nil { logit.Error.Println("error when trying to remove disk volume" + err.Error()) } } i++ } //delete the container entries //delete the cluster entry admindb.DeleteCluster(dbConn, ID) for i = range containers { err = admindb.DeleteContainer(dbConn, containers[i].ID) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } } status := types.SimpleStatus{} status.Status = "OK" w.WriteHeader(http.StatusOK) w.WriteJson(&status) }
// ScaleUpCluster increases the count of standby containers in a cluster func ScaleUpCluster(w rest.ResponseWriter, r *rest.Request) { dbConn, err := util.GetConnection(CLUSTERADMIN_DB) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), 400) return } defer dbConn.Close() err = secimpl.Authorize(dbConn, r.PathParam("Token"), "perm-read") if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusUnauthorized) return } ID := r.PathParam("ID") cluster, err := admindb.GetCluster(dbConn, ID) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } var containers []types.Container containers, err = admindb.GetAllContainersForCluster(dbConn, ID) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } //determine number of standby nodes currently standbyCnt := 0 for i := range containers { if containers[i].Role == STANDBY { standbyCnt++ } } //logit.Info.Printf("standbyCnt ends at %d\n", standbyCnt) //provision new container params := new(swarmapi.DockerRunRequest) params.Image = "cpm-node" //TODO make the server choice smart params.ProjectID = cluster.ProjectID params.ContainerName = cluster.Name + "-" + STANDBY + "-" + fmt.Sprintf("%d", standbyCnt) params.Standalone = "false" var standby = true params.Profile = "LG" //logit.Info.Printf("here with ProjectID %s\n", cluster.ProjectID) _, err = provisionImpl(dbConn, params, standby) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } err = provisionImplInit(dbConn, params, false) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } //need to update the new container's ClusterID var node types.Container node, err = admindb.GetContainerByName(dbConn, params.ContainerName) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, "error"+err.Error(), http.StatusBadRequest) return } node.ClusterID = cluster.ID node.Role = STANDBY err = admindb.UpdateContainer(dbConn, node) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, "error"+err.Error(), http.StatusBadRequest) return } err = configureCluster(params.Profile, dbConn, cluster, false) if err != nil { logit.Error.Println(err.Error()) rest.Error(w, err.Error(), http.StatusBadRequest) return } w.WriteHeader(http.StatusOK) status := types.SimpleStatus{} status.Status = "OK" w.WriteJson(&status) }