use of com.cloud.context.CallContext in project cosmic by MissionCriticalCloud.
the class VirtualMachineManagerImpl method orchestrateStart.
@Override
public void orchestrateStart(final String vmUuid, final Map<VirtualMachineProfile.Param, Object> params, final DeploymentPlan planToDeploy, final DeploymentPlanner planner) throws InsufficientCapacityException, ConcurrentOperationException, ResourceUnavailableException {
final CallContext cctxt = CallContext.current();
final Account account = cctxt.getCallingAccount();
final User caller = cctxt.getCallingUser();
VMInstanceVO vm = _vmDao.findByUuid(vmUuid);
final Ternary<VMInstanceVO, ReservationContext, ItWorkVO> start = changeToStartState(vm, caller, account);
if (start == null) {
return;
}
vm = start.first();
final ReservationContext ctx = start.second();
ItWorkVO work = start.third();
VMInstanceVO startedVm = null;
final ServiceOfferingVO offering = _offeringDao.findById(vm.getId(), vm.getServiceOfferingId());
final VirtualMachineTemplate template = _entityMgr.findByIdIncludingRemoved(VirtualMachineTemplate.class, vm.getTemplateId());
s_logger.debug("Trying to deploy VM, vm has dcId: " + vm.getDataCenterId() + " and podId: " + vm.getPodIdToDeployIn());
DataCenterDeployment plan = new DataCenterDeployment(vm.getDataCenterId(), vm.getPodIdToDeployIn(), null, null, null, null, ctx);
if (planToDeploy != null && planToDeploy.getDataCenterId() != 0) {
s_logger.debug("advanceStart: DeploymentPlan is provided, using dcId:" + planToDeploy.getDataCenterId() + ", podId: " + planToDeploy.getPodId() + ", clusterId: " + planToDeploy.getClusterId() + ", hostId: " + planToDeploy.getHostId() + ", poolId: " + planToDeploy.getPoolId());
plan = new DataCenterDeployment(planToDeploy.getDataCenterId(), planToDeploy.getPodId(), planToDeploy.getClusterId(), planToDeploy.getHostId(), planToDeploy.getPoolId(), planToDeploy.getPhysicalNetworkId(), ctx);
}
final HypervisorGuru hvGuru = _hvGuruMgr.getGuru(vm.getHypervisorType());
final VirtualMachineGuru vmGuru = getVmGuru(vm);
boolean canRetry = true;
ExcludeList avoids = null;
try {
final Journal journal = start.second().getJournal();
if (planToDeploy != null) {
avoids = planToDeploy.getAvoids();
}
if (avoids == null) {
avoids = new ExcludeList();
}
s_logger.debug("VM start orchestration will {}", avoids.toString());
boolean planChangedByVolume = false;
boolean reuseVolume = true;
final DataCenterDeployment originalPlan = plan;
int retry = StartRetry.value();
while (retry-- != 0) {
if (reuseVolume) {
// edit plan if this vm's ROOT volume is in READY state already
final List<VolumeVO> vols = _volsDao.findReadyRootVolumesByInstance(vm.getId());
for (final VolumeVO vol : vols) {
// make sure if the templateId is unchanged. If it is changed,
// let planner
// reassign pool for the volume even if it ready.
final Long volTemplateId = vol.getTemplateId();
if (volTemplateId != null && volTemplateId != template.getId()) {
s_logger.debug(vol + " of " + vm + " is READY, but template ids don't match, let the planner reassign a new pool");
continue;
}
final StoragePool pool = (StoragePool) dataStoreMgr.getPrimaryDataStore(vol.getPoolId());
if (!pool.isInMaintenance()) {
s_logger.debug("Root volume is ready, need to place VM in volume's cluster");
final long rootVolDcId = pool.getDataCenterId();
final Long rootVolPodId = pool.getPodId();
final Long rootVolClusterId = pool.getClusterId();
if (planToDeploy != null && planToDeploy.getDataCenterId() != 0) {
final Long clusterIdSpecified = planToDeploy.getClusterId();
if (clusterIdSpecified != null && rootVolClusterId != null) {
checkIfPlanIsDeployable(vm, rootVolClusterId, clusterIdSpecified);
}
plan = new DataCenterDeployment(planToDeploy.getDataCenterId(), planToDeploy.getPodId(), planToDeploy.getClusterId(), planToDeploy.getHostId(), vol.getPoolId(), null, ctx);
} else {
plan = new DataCenterDeployment(rootVolDcId, rootVolPodId, rootVolClusterId, null, vol.getPoolId(), null, ctx);
s_logger.debug(vol + " is READY, changing deployment plan to use this pool's dcId: " + rootVolDcId + " , podId: " + rootVolPodId + " , and clusterId: " + rootVolClusterId);
planChangedByVolume = true;
}
}
}
}
final Account owner = _entityMgr.findById(Account.class, vm.getAccountId());
final VirtualMachineProfileImpl vmProfile = new VirtualMachineProfileImpl(vm, template, offering, owner, params);
DeployDestination dest;
try {
dest = _dpMgr.planDeployment(vmProfile, plan, avoids, planner);
} catch (final AffinityConflictException e2) {
s_logger.warn("Unable to create deployment, affinity rules associted to the VM conflict", e2);
throw new CloudRuntimeException("Unable to create deployment, affinity rules associted to the VM conflict");
}
if (dest == null) {
if (planChangedByVolume) {
plan = originalPlan;
planChangedByVolume = false;
// do not enter volume reuse for next retry, since we want to look for resources outside the volume's cluster
reuseVolume = false;
continue;
}
throw new InsufficientServerCapacityException("Unable to create a deployment for " + vmProfile, DataCenter.class, plan.getDataCenterId(), areAffinityGroupsAssociated(vmProfile));
}
if (dest != null) {
avoids.addHost(dest.getHost().getId());
journal.record("Deployment found ", vmProfile, dest);
}
long destHostId = dest.getHost().getId();
vm.setPodIdToDeployIn(dest.getPod().getId());
final Long cluster_id = dest.getCluster().getId();
final ClusterDetailsVO cluster_detail_cpu = _clusterDetailsDao.findDetail(cluster_id, "cpuOvercommitRatio");
final ClusterDetailsVO cluster_detail_ram = _clusterDetailsDao.findDetail(cluster_id, "memoryOvercommitRatio");
// storing the value of overcommit in the vm_details table for doing a capacity check in case the cluster overcommit ratio is changed.
if (_uservmDetailsDao.findDetail(vm.getId(), "cpuOvercommitRatio") == null && (Float.parseFloat(cluster_detail_cpu.getValue()) > 1f || Float.parseFloat(cluster_detail_ram.getValue()) > 1f)) {
_uservmDetailsDao.addDetail(vm.getId(), "cpuOvercommitRatio", cluster_detail_cpu.getValue(), true);
_uservmDetailsDao.addDetail(vm.getId(), "memoryOvercommitRatio", cluster_detail_ram.getValue(), true);
} else if (_uservmDetailsDao.findDetail(vm.getId(), "cpuOvercommitRatio") != null) {
_uservmDetailsDao.addDetail(vm.getId(), "cpuOvercommitRatio", cluster_detail_cpu.getValue(), true);
_uservmDetailsDao.addDetail(vm.getId(), "memoryOvercommitRatio", cluster_detail_ram.getValue(), true);
}
vmProfile.setCpuOvercommitRatio(Float.parseFloat(cluster_detail_cpu.getValue()));
vmProfile.setMemoryOvercommitRatio(Float.parseFloat(cluster_detail_ram.getValue()));
StartAnswer startAnswer = null;
try {
if (!changeState(vm, Event.OperationRetry, destHostId, work, Step.Prepare)) {
throw new ConcurrentOperationException("Unable to update the state of the Virtual Machine " + vm.getUuid() + " oldstate: " + vm.getState() + "Event :" + Event.OperationRetry);
}
} catch (final NoTransitionException e1) {
throw new ConcurrentOperationException(e1.getMessage());
}
try {
s_logger.debug("VM is being created in podId: " + vm.getPodIdToDeployIn());
_networkMgr.prepare(vmProfile, dest, ctx);
volumeMgr.prepare(vmProfile, dest);
// since StorageMgr succeeded in volume creation, reuse Volume for further tries until current cluster has capacity
if (!reuseVolume) {
reuseVolume = true;
}
vmGuru.finalizeVirtualMachineProfile(vmProfile, dest, ctx);
final VirtualMachineTO vmTO = hvGuru.implement(vmProfile);
handlePath(vmTO.getDisks(), vm.getHypervisorType());
final Commands cmds = new Commands(Command.OnError.Stop);
cmds.addCommand(new StartCommand(vmTO, dest.getHost(), getExecuteInSequence(vm.getHypervisorType())));
vmGuru.finalizeDeployment(cmds, vmProfile, dest, ctx);
work = _workDao.findById(work.getId());
if (work == null || work.getStep() != Step.Prepare) {
throw new ConcurrentOperationException("Work steps have been changed: " + work);
}
_workDao.updateStep(work, Step.Starting);
_agentMgr.send(destHostId, cmds);
_workDao.updateStep(work, Step.Started);
startAnswer = cmds.getAnswer(StartAnswer.class);
if (startAnswer != null && startAnswer.getResult()) {
handlePath(vmTO.getDisks(), startAnswer.getIqnToPath());
final String host_guid = startAnswer.getHost_guid();
if (host_guid != null) {
final HostVO finalHost = _resourceMgr.findHostByGuid(host_guid);
if (finalHost == null) {
throw new CloudRuntimeException("Host Guid " + host_guid + " doesn't exist in DB, something went wrong while processing start answer: " + startAnswer);
}
destHostId = finalHost.getId();
}
if (vmGuru.finalizeStart(vmProfile, destHostId, cmds, ctx)) {
syncDiskChainChange(startAnswer);
if (!changeState(vm, Event.OperationSucceeded, destHostId, work, Step.Done)) {
s_logger.error("Unable to transition to a new state. VM uuid: " + vm.getUuid() + "VM oldstate:" + vm.getState() + "Event:" + Event.OperationSucceeded);
throw new ConcurrentOperationException("Failed to deploy VM" + vm.getUuid());
}
// Update GPU device capacity
final GPUDeviceTO gpuDevice = startAnswer.getVirtualMachine().getGpuDevice();
if (gpuDevice != null) {
_resourceMgr.updateGPUDetails(destHostId, gpuDevice.getGroupDetails());
}
startedVm = vm;
s_logger.debug("Start completed for VM " + vm);
return;
} else {
s_logger.info("The guru did not like the answers so stopping " + vm);
final StopCommand cmd = new StopCommand(vm, getExecuteInSequence(vm.getHypervisorType()), false);
final Answer answer = _agentMgr.easySend(destHostId, cmd);
if (answer != null && answer instanceof StopAnswer) {
final StopAnswer stopAns = (StopAnswer) answer;
if (vm.getType() == VirtualMachine.Type.User) {
final String platform = stopAns.getPlatform();
if (platform != null) {
final Map<String, String> vmmetadata = new HashMap<>();
vmmetadata.put(vm.getInstanceName(), platform);
syncVMMetaData(vmmetadata);
}
}
}
if (answer == null || !answer.getResult()) {
s_logger.warn("Unable to stop " + vm + " due to " + (answer != null ? answer.getDetails() : "no answers"));
_haMgr.scheduleStop(vm, destHostId, WorkType.ForceStop);
throw new ExecutionException("Unable to stop this VM, " + vm.getUuid() + " so we are unable to retry the start operation");
}
throw new ExecutionException("Unable to start VM:" + vm.getUuid() + " due to error in finalizeStart, not retrying");
}
}
s_logger.info("Unable to start VM on " + dest.getHost() + " due to " + (startAnswer == null ? " no start answer" : startAnswer.getDetails()));
if (startAnswer != null && startAnswer.getContextParam("stopRetry") != null) {
break;
}
} catch (final OperationTimedoutException e) {
s_logger.debug("Unable to send the start command to host " + dest.getHost() + " failed to start VM: " + vm.getUuid());
if (e.isActive()) {
_haMgr.scheduleStop(vm, destHostId, WorkType.CheckStop);
}
canRetry = false;
throw new AgentUnavailableException("Unable to start " + vm.getHostName(), destHostId, e);
} catch (final ResourceUnavailableException e) {
s_logger.info("Unable to contact resource.", e);
if (!avoids.add(e)) {
if (e.getScope() == Volume.class || e.getScope() == Nic.class) {
throw e;
} else {
s_logger.warn("unexpected ResourceUnavailableException : " + e.getScope().getName(), e);
throw e;
}
}
} catch (final InsufficientCapacityException e) {
s_logger.info("Insufficient capacity ", e);
if (!avoids.add(e)) {
if (e.getScope() == Volume.class || e.getScope() == Nic.class) {
throw e;
} else {
s_logger.warn("unexpected InsufficientCapacityException : " + e.getScope().getName(), e);
}
}
} catch (final ExecutionException | NoTransitionException e) {
s_logger.error("Failed to start instance " + vm, e);
throw new AgentUnavailableException("Unable to start instance due to " + e.getMessage(), destHostId, e);
} finally {
if (startedVm == null && canRetry) {
final Step prevStep = work.getStep();
_workDao.updateStep(work, Step.Release);
// If previous step was started/ing && we got a valid answer
if ((prevStep == Step.Started || prevStep == Step.Starting) && startAnswer != null && startAnswer.getResult()) {
// TODO check the response of cleanup
// and record it in DB for retry
cleanup(vmGuru, vmProfile, work, Event.OperationFailed, false);
} else {
// if step is not starting/started, send cleanup command with force=true
cleanup(vmGuru, vmProfile, work, Event.OperationFailed, true);
}
}
}
}
} finally {
if (startedVm == null) {
if (canRetry) {
try {
changeState(vm, Event.OperationFailed, null, work, Step.Done);
} catch (final NoTransitionException e) {
throw new ConcurrentOperationException(e.getMessage());
}
}
}
if (planToDeploy != null) {
planToDeploy.setAvoids(avoids);
}
}
if (startedVm == null) {
throw new CloudRuntimeException("Unable to start instance '" + vm.getHostName() + "' (" + vm.getUuid() + "), see management server log for details");
}
}
use of com.cloud.context.CallContext in project cosmic by MissionCriticalCloud.
the class VirtualMachineManagerImpl method startVmThroughJobQueue.
//
// PowerState report handling for out-of-band changes and handling of left-over transitional VM states
//
//
// VM operation based on new sync model
//
//
// TODO build a common pattern to reduce code duplication in following methods
// no time for this at current iteration
//
public Outcome<VirtualMachine> startVmThroughJobQueue(final String vmUuid, final Map<VirtualMachineProfile.Param, Object> params, final DeploymentPlan planToDeploy, final DeploymentPlanner planner) {
final CallContext context = CallContext.current();
final User callingUser = context.getCallingUser();
final Account callingAccount = context.getCallingAccount();
final VMInstanceVO vm = _vmDao.findByUuid(vmUuid);
VmWorkJobVO workJob = null;
final List<VmWorkJobVO> pendingWorkJobs = _workJobDao.listPendingWorkJobs(VirtualMachine.Type.Instance, vm.getId(), VmWorkStart.class.getName());
if (pendingWorkJobs.size() > 0) {
assert pendingWorkJobs.size() == 1;
workJob = pendingWorkJobs.get(0);
} else {
workJob = new VmWorkJobVO(context.getContextId());
workJob.setDispatcher(VmWorkConstants.VM_WORK_JOB_DISPATCHER);
workJob.setCmd(VmWorkStart.class.getName());
workJob.setAccountId(callingAccount.getId());
workJob.setUserId(callingUser.getId());
workJob.setStep(VmWorkJobVO.Step.Starting);
workJob.setVmType(VirtualMachine.Type.Instance);
workJob.setVmInstanceId(vm.getId());
workJob.setRelated(AsyncJobExecutionContext.getOriginJobId());
// save work context info (there are some duplications)
final VmWorkStart workInfo = new VmWorkStart(callingUser.getId(), callingAccount.getId(), vm.getId(), VirtualMachineManagerImpl.VM_WORK_JOB_HANDLER);
workInfo.setPlan(planToDeploy);
workInfo.setParams(params);
if (planner != null) {
workInfo.setDeploymentPlanner(planner.getName());
}
workJob.setCmdInfo(VmWorkSerializer.serialize(workInfo));
_jobMgr.submitAsyncJob(workJob, VmWorkConstants.VM_WORK_QUEUE, vm.getId());
}
AsyncJobExecutionContext.getCurrentExecutionContext().joinJob(workJob.getId());
return new VmStateSyncOutcome(workJob, VirtualMachine.PowerState.PowerOn, vm.getId(), null);
}
use of com.cloud.context.CallContext in project cosmic by MissionCriticalCloud.
the class VirtualMachineManagerImpl method migrateVmAwayThroughJobQueue.
public Outcome<VirtualMachine> migrateVmAwayThroughJobQueue(final String vmUuid, final long srcHostId) {
final CallContext context = CallContext.current();
final User user = context.getCallingUser();
final Account account = context.getCallingAccount();
final VMInstanceVO vm = _vmDao.findByUuid(vmUuid);
final List<VmWorkJobVO> pendingWorkJobs = _workJobDao.listPendingWorkJobs(VirtualMachine.Type.Instance, vm.getId(), VmWorkMigrateAway.class.getName());
VmWorkJobVO workJob = null;
if (pendingWorkJobs != null && pendingWorkJobs.size() > 0) {
assert pendingWorkJobs.size() == 1;
workJob = pendingWorkJobs.get(0);
} else {
workJob = new VmWorkJobVO(context.getContextId());
workJob.setDispatcher(VmWorkConstants.VM_WORK_JOB_DISPATCHER);
workJob.setCmd(VmWorkMigrateAway.class.getName());
workJob.setAccountId(account.getId());
workJob.setUserId(user.getId());
workJob.setVmType(VirtualMachine.Type.Instance);
workJob.setVmInstanceId(vm.getId());
workJob.setRelated(AsyncJobExecutionContext.getOriginJobId());
// save work context info (there are some duplications)
final VmWorkMigrateAway workInfo = new VmWorkMigrateAway(user.getId(), account.getId(), vm.getId(), VirtualMachineManagerImpl.VM_WORK_JOB_HANDLER, srcHostId);
workJob.setCmdInfo(VmWorkSerializer.serialize(workInfo));
}
_jobMgr.submitAsyncJob(workJob, VmWorkConstants.VM_WORK_QUEUE, vm.getId());
AsyncJobExecutionContext.getCurrentExecutionContext().joinJob(workJob.getId());
return new VmStateSyncOutcome(workJob, VirtualMachine.PowerState.PowerOn, vm.getId(), vm.getPowerHostId());
}
use of com.cloud.context.CallContext in project cosmic by MissionCriticalCloud.
the class VirtualMachineManagerImpl method orchestrateRemoveVmFromNetwork.
@DB
private boolean orchestrateRemoveVmFromNetwork(final VirtualMachine vm, final Network network, final URI broadcastUri) throws ConcurrentOperationException, ResourceUnavailableException {
final CallContext cctx = CallContext.current();
final VMInstanceVO vmVO = _vmDao.findById(vm.getId());
final ReservationContext context = new ReservationContextImpl(null, null, cctx.getCallingUser(), cctx.getCallingAccount());
final VirtualMachineProfileImpl vmProfile = new VirtualMachineProfileImpl(vmVO, null, null, null, null);
final Zone zone = _zoneRepository.findOne(network.getDataCenterId());
final Host host = _hostDao.findById(vm.getHostId());
final DeployDestination dest = new DeployDestination(zone, null, null, host);
final HypervisorGuru hvGuru = _hvGuruMgr.getGuru(vmProfile.getVirtualMachine().getHypervisorType());
final VirtualMachineTO vmTO = hvGuru.implement(vmProfile);
Nic nic = null;
if (broadcastUri != null) {
nic = _nicsDao.findByNetworkIdInstanceIdAndBroadcastUri(network.getId(), vm.getId(), broadcastUri.toString());
} else {
nic = _networkModel.getNicInNetwork(vm.getId(), network.getId());
}
if (nic == null) {
s_logger.warn("Could not get a nic with " + network);
return false;
}
// don't delete default NIC on a user VM
if (nic.isDefaultNic() && vm.getType() == VirtualMachine.Type.User) {
s_logger.warn("Failed to remove nic from " + vm + " in " + network + ", nic is default.");
throw new CloudRuntimeException("Failed to remove nic from " + vm + " in " + network + ", nic is default.");
}
// Lock on nic is needed here
final Nic lock = _nicsDao.acquireInLockTable(nic.getId());
if (lock == null) {
// check if nic is still there. Return if it was released already
if (_nicsDao.findById(nic.getId()) == null) {
s_logger.debug("Not need to remove the vm " + vm + " from network " + network + " as the vm doesn't have nic in this network");
return true;
}
throw new ConcurrentOperationException("Unable to lock nic " + nic.getId());
}
s_logger.debug("Lock is acquired for nic id " + lock.getId() + " as a part of remove vm " + vm + " from network " + network);
try {
final NicProfile nicProfile = new NicProfile(nic, network, nic.getBroadcastUri(), nic.getIsolationUri(), _networkModel.getNetworkRate(network.getId(), vm.getId()), _networkModel.getNetworkTag(vmProfile.getVirtualMachine().getHypervisorType(), network));
// 1) Unplug the nic
if (vm.getState() == State.Running) {
final NicTO nicTO = toNicTO(nicProfile, vmProfile.getVirtualMachine().getHypervisorType());
s_logger.debug("Un-plugging nic for vm " + vm + " from network " + network);
final boolean result = unplugNic(network, nicTO, vmTO, context, dest);
if (result) {
s_logger.debug("Nic is unplugged successfully for vm " + vm + " in network " + network);
} else {
s_logger.warn("Failed to unplug nic for the vm " + vm + " from network " + network);
return false;
}
} else if (vm.getState() != State.Stopped) {
s_logger.warn("Unable to remove vm " + vm + " from network " + network);
throw new ResourceUnavailableException("Unable to remove vm " + vm + " from network, is not in the right state", DataCenter.class, vm.getDataCenterId());
}
// 2) Release the nic
_networkMgr.releaseNic(vmProfile, nic);
s_logger.debug("Successfully released nic " + nic + "for vm " + vm);
// 3) Remove the nic
_networkMgr.removeNic(vmProfile, nic);
return true;
} finally {
if (lock != null) {
_nicsDao.releaseFromLockTable(lock.getId());
s_logger.debug("Lock is released for nic id " + lock.getId() + " as a part of remove vm " + vm + " from network " + network);
}
}
}
use of com.cloud.context.CallContext in project cosmic by MissionCriticalCloud.
the class VirtualMachineManagerImpl method removeVmFromNetworkThroughJobQueue.
public Outcome<VirtualMachine> removeVmFromNetworkThroughJobQueue(final VirtualMachine vm, final Network network, final URI broadcastUri) {
final CallContext context = CallContext.current();
final User user = context.getCallingUser();
final Account account = context.getCallingAccount();
final List<VmWorkJobVO> pendingWorkJobs = _workJobDao.listPendingWorkJobs(VirtualMachine.Type.Instance, vm.getId(), VmWorkRemoveVmFromNetwork.class.getName());
VmWorkJobVO workJob = null;
if (pendingWorkJobs != null && pendingWorkJobs.size() > 0) {
assert pendingWorkJobs.size() == 1;
workJob = pendingWorkJobs.get(0);
} else {
workJob = new VmWorkJobVO(context.getContextId());
workJob.setDispatcher(VmWorkConstants.VM_WORK_JOB_DISPATCHER);
workJob.setCmd(VmWorkRemoveVmFromNetwork.class.getName());
workJob.setAccountId(account.getId());
workJob.setUserId(user.getId());
workJob.setVmType(VirtualMachine.Type.Instance);
workJob.setVmInstanceId(vm.getId());
workJob.setRelated(AsyncJobExecutionContext.getOriginJobId());
// save work context info (there are some duplications)
final VmWorkRemoveVmFromNetwork workInfo = new VmWorkRemoveVmFromNetwork(user.getId(), account.getId(), vm.getId(), VirtualMachineManagerImpl.VM_WORK_JOB_HANDLER, network, broadcastUri);
workJob.setCmdInfo(VmWorkSerializer.serialize(workInfo));
_jobMgr.submitAsyncJob(workJob, VmWorkConstants.VM_WORK_QUEUE, vm.getId());
}
AsyncJobExecutionContext.getCurrentExecutionContext().joinJob(workJob.getId());
return new VmJobVirtualMachineOutcome(workJob, vm.getId());
}
Aggregations