@Test
public void testPoolStateIsNotUp() {
try {
createDb();
StoragePoolVO pool = storagePoolDao.findById(storagePoolId);
pool.setScope(ScopeType.ZONE);
pool.setStatus(StoragePoolStatus.Maintenance);
storagePoolDao.update(pool.getId(), pool);
DiskProfile profile = new DiskProfile(volume, diskOffering, HypervisorType.XenServer);
VirtualMachineProfile vmProfile = Mockito.mock(VirtualMachineProfile.class);
Mockito.when(
storageMgr.storagePoolHasEnoughSpace(
Mockito.anyListOf(Volume.class), Mockito.any(StoragePool.class)))
.thenReturn(true);
DeploymentPlan plan = new DataCenterDeployment(dcId, podId, clusterId, null, null, null);
int foundAcct = 0;
for (StoragePoolAllocator allocator : allocators) {
List<StoragePool> pools =
allocator.allocateToPool(profile, vmProfile, plan, new ExcludeList(), 1);
if (!pools.isEmpty()) {
Assert.assertEquals(pools.get(0).getId(), storage.getId());
foundAcct++;
}
}
if (foundAcct == 1) {
Assert.fail();
}
} catch (Exception e) {
cleanDb();
Assert.fail();
}
}
@Override
public boolean attachCluster(DataStore store, ClusterScope scope) {
PrimaryDataStoreInfo primarystore = (PrimaryDataStoreInfo) store;
// Check if there is host up in this cluster
List<HostVO> allHosts =
_resourceMgr.listAllUpAndEnabledHosts(
Host.Type.Routing,
primarystore.getClusterId(),
primarystore.getPodId(),
primarystore.getDataCenterId());
if (allHosts.isEmpty()) {
primaryDataStoreDao.expunge(primarystore.getId());
throw new CloudRuntimeException(
"No host up to associate a storage pool with in cluster " + primarystore.getClusterId());
}
if (primarystore.getPoolType() == StoragePoolType.OCFS2
&& !_ocfs2Mgr.prepareNodes(allHosts, primarystore)) {
s_logger.warn(
"Can not create storage pool "
+ primarystore
+ " on cluster "
+ primarystore.getClusterId());
primaryDataStoreDao.expunge(primarystore.getId());
return false;
}
boolean success = false;
for (HostVO h : allHosts) {
success = createStoragePool(h.getId(), primarystore);
if (success) {
break;
}
}
s_logger.debug("In createPool Adding the pool to each of the hosts");
List<HostVO> poolHosts = new ArrayList<HostVO>();
for (HostVO h : allHosts) {
try {
storageMgr.connectHostToSharedPool(h.getId(), primarystore.getId());
poolHosts.add(h);
} catch (Exception e) {
s_logger.warn("Unable to establish a connection between " + h + " and " + primarystore, e);
}
}
if (poolHosts.isEmpty()) {
s_logger.warn(
"No host can access storage pool "
+ primarystore
+ " on cluster "
+ primarystore.getClusterId());
primaryDataStoreDao.expunge(primarystore.getId());
throw new CloudRuntimeException("Failed to access storage pool");
}
dataStoreHelper.attachCluster(store);
return true;
}
private SummedCapacity getUsedStats(short capacityType, long zoneId, Long podId, Long clusterId) {
CapacityVO capacity;
if (capacityType == Capacity.CAPACITY_TYPE_SECONDARY_STORAGE) {
capacity = _storageMgr.getSecondaryStorageUsedStats(null, zoneId);
} else {
capacity = _storageMgr.getStoragePoolUsedStats(null, clusterId, podId, zoneId);
}
if (capacity != null) {
return new SummedCapacity(
capacity.getUsedCapacity(),
0,
capacity.getTotalCapacity(),
capacityType,
clusterId,
podId);
} else {
return null;
}
}
@Test
public void testClusterAllocatorMultiplePools() {
Long newStorageId = null;
try {
createDb();
DataStoreProvider provider =
providerMgr.getDataStoreProvider("ancient primary data store provider");
storage = new StoragePoolVO();
storage.setDataCenterId(dcId);
storage.setPodId(podId);
storage.setPoolType(StoragePoolType.NetworkFilesystem);
storage.setClusterId(clusterId);
storage.setStatus(StoragePoolStatus.Up);
storage.setScope(ScopeType.CLUSTER);
storage.setAvailableBytes(1000);
storage.setCapacityBytes(20000);
storage.setHostAddress(UUID.randomUUID().toString());
storage.setPath(UUID.randomUUID().toString());
storage.setStorageProviderName(provider.getName());
StoragePoolVO newStorage = storagePoolDao.persist(storage);
newStorageId = newStorage.getId();
DiskProfile profile = new DiskProfile(volume, diskOffering, HypervisorType.XenServer);
VirtualMachineProfile vmProfile = Mockito.mock(VirtualMachineProfile.class);
Mockito.when(
storageMgr.storagePoolHasEnoughSpace(
Mockito.anyListOf(Volume.class), Mockito.any(StoragePool.class)))
.thenReturn(true);
DeploymentPlan plan = new DataCenterDeployment(dcId, podId, clusterId, null, null, null);
int foundAcct = 0;
for (StoragePoolAllocator allocator : allocators) {
List<StoragePool> pools =
allocator.allocateToPool(profile, vmProfile, plan, new ExcludeList(), 1);
if (!pools.isEmpty()) {
Assert.assertEquals(pools.size(), 1);
foundAcct++;
}
}
if (foundAcct > 1 || foundAcct == 0) {
Assert.fail();
}
} catch (Exception e) {
cleanDb();
if (newStorageId != null) {
storagePoolDao.remove(newStorageId);
}
Assert.fail();
}
}
@Test
public void testClusterAllocatorWithTags() {
try {
createDb();
StoragePoolDetailVO detailVO = new StoragePoolDetailVO(this.storagePoolId, "high", "true");
poolDetailsDao.persist(detailVO);
DiskOfferingVO diskOff = this.diskOfferingDao.findById(diskOffering.getId());
List<String> tags = new ArrayList<String>();
tags.add("high");
diskOff.setTagsArray(tags);
diskOfferingDao.update(diskOff.getId(), diskOff);
DiskProfile profile = new DiskProfile(volume, diskOff, HypervisorType.XenServer);
VirtualMachineProfile vmProfile = Mockito.mock(VirtualMachineProfile.class);
Mockito.when(
storageMgr.storagePoolHasEnoughSpace(
Mockito.anyListOf(Volume.class), Mockito.any(StoragePool.class)))
.thenReturn(true);
DeploymentPlan plan = new DataCenterDeployment(dcId, podId, clusterId, null, null, null);
int foundAcct = 0;
for (StoragePoolAllocator allocator : allocators) {
List<StoragePool> pools =
allocator.allocateToPool(profile, vmProfile, plan, new ExcludeList(), 1);
if (!pools.isEmpty()) {
Assert.assertEquals(pools.get(0).getId(), storage.getId());
foundAcct++;
}
}
if (foundAcct > 1 || foundAcct == 0) {
Assert.fail();
}
} catch (Exception e) {
cleanDb();
Assert.fail();
}
}
protected Long restart(HaWorkVO work) {
List<HaWorkVO> items = _haDao.listFutureHaWorkForVm(work.getInstanceId(), work.getId());
if (items.size() > 0) {
StringBuilder str =
new StringBuilder(
"Cancelling this work item because newer ones have been scheduled. Work Ids = [");
for (HaWorkVO item : items) {
str.append(item.getId()).append(", ");
}
str.delete(str.length() - 2, str.length()).append("]");
s_logger.info(str.toString());
return null;
}
items = _haDao.listRunningHaWorkForVm(work.getInstanceId());
if (items.size() > 0) {
StringBuilder str =
new StringBuilder(
"Waiting because there's HA work being executed on an item currently. Work Ids =[");
for (HaWorkVO item : items) {
str.append(item.getId()).append(", ");
}
str.delete(str.length() - 2, str.length()).append("]");
s_logger.info(str.toString());
return (System.currentTimeMillis() >> 10) + _investigateRetryInterval;
}
long vmId = work.getInstanceId();
VMInstanceVO vm = _itMgr.findByIdAndType(work.getType(), work.getInstanceId());
if (vm == null) {
s_logger.info("Unable to find vm: " + vmId);
return null;
}
s_logger.info("HA on " + vm);
if (vm.getState() != work.getPreviousState() || vm.getUpdated() != work.getUpdateTime()) {
s_logger.info(
"VM "
+ vm
+ " has been changed. Current State = "
+ vm.getState()
+ " Previous State = "
+ work.getPreviousState()
+ " last updated = "
+ vm.getUpdated()
+ " previous updated = "
+ work.getUpdateTime());
return null;
}
short alertType = AlertManager.ALERT_TYPE_USERVM;
if (VirtualMachine.Type.DomainRouter.equals(vm.getType())) {
alertType = AlertManager.ALERT_TYPE_DOMAIN_ROUTER;
} else if (VirtualMachine.Type.ConsoleProxy.equals(vm.getType())) {
alertType = AlertManager.ALERT_TYPE_CONSOLE_PROXY;
} else if (VirtualMachine.Type.SecondaryStorageVm.equals(vm.getType())) {
alertType = AlertManager.ALERT_TYPE_SSVM;
}
HostVO host = _hostDao.findById(work.getHostId());
boolean isHostRemoved = false;
if (host == null) {
host = _hostDao.findByIdIncludingRemoved(work.getHostId());
if (host != null) {
s_logger.debug(
"VM "
+ vm.toString()
+ " is now no longer on host "
+ work.getHostId()
+ " as the host is removed");
isHostRemoved = true;
}
}
DataCenterVO dcVO = _dcDao.findById(host.getDataCenterId());
HostPodVO podVO = _podDao.findById(host.getPodId());
String hostDesc =
"name: "
+ host.getName()
+ "(id:"
+ host.getId()
+ "), availability zone: "
+ dcVO.getName()
+ ", pod: "
+ podVO.getName();
Boolean alive = null;
if (work.getStep() == Step.Investigating) {
if (!isHostRemoved) {
if (vm.getHostId() == null || vm.getHostId() != work.getHostId()) {
s_logger.info("VM " + vm.toString() + " is now no longer on host " + work.getHostId());
return null;
}
Enumeration<Investigator> en = _investigators.enumeration();
Investigator investigator = null;
while (en.hasMoreElements()) {
investigator = en.nextElement();
alive = investigator.isVmAlive(vm, host);
s_logger.info(investigator.getName() + " found " + vm + "to be alive? " + alive);
if (alive != null) {
break;
}
}
boolean fenced = false;
if (alive == null) {
s_logger.debug("Fencing off VM that we don't know the state of");
Enumeration<FenceBuilder> enfb = _fenceBuilders.enumeration();
while (enfb.hasMoreElements()) {
FenceBuilder fb = enfb.nextElement();
Boolean result = fb.fenceOff(vm, host);
s_logger.info("Fencer " + fb.getName() + " returned " + result);
if (result != null && result) {
fenced = true;
break;
}
}
} else if (!alive) {
fenced = true;
} else {
s_logger.debug(
"VM " + vm.getHostName() + " is found to be alive by " + investigator.getName());
if (host.getStatus() == Status.Up) {
s_logger.info(vm + " is alive and host is up. No need to restart it.");
return null;
} else {
s_logger.debug("Rescheduling because the host is not up but the vm is alive");
return (System.currentTimeMillis() >> 10) + _investigateRetryInterval;
}
}
if (!fenced) {
s_logger.debug("We were unable to fence off the VM " + vm);
_alertMgr.sendAlert(
alertType,
vm.getDataCenterIdToDeployIn(),
vm.getPodIdToDeployIn(),
"Unable to restart " + vm.getHostName() + " which was running on host " + hostDesc,
"Insufficient capacity to restart VM, name: "
+ vm.getHostName()
+ ", id: "
+ vmId
+ " which was running on host "
+ hostDesc);
return (System.currentTimeMillis() >> 10) + _restartRetryInterval;
}
try {
_itMgr.advanceStop(vm, true, _accountMgr.getSystemUser(), _accountMgr.getSystemAccount());
} catch (ResourceUnavailableException e) {
assert false : "How do we hit this when force is true?";
throw new CloudRuntimeException("Caught exception even though it should be handled.", e);
} catch (OperationTimedoutException e) {
assert false : "How do we hit this when force is true?";
throw new CloudRuntimeException("Caught exception even though it should be handled.", e);
} catch (ConcurrentOperationException e) {
assert false : "How do we hit this when force is true?";
throw new CloudRuntimeException("Caught exception even though it should be handled.", e);
}
work.setStep(Step.Scheduled);
_haDao.update(work.getId(), work);
} else {
s_logger.debug(
"How come that HA step is Investigating and the host is removed? Calling forced Stop on Vm anyways");
try {
_itMgr.advanceStop(vm, true, _accountMgr.getSystemUser(), _accountMgr.getSystemAccount());
} catch (ResourceUnavailableException e) {
assert false : "How do we hit this when force is true?";
throw new CloudRuntimeException("Caught exception even though it should be handled.", e);
} catch (OperationTimedoutException e) {
assert false : "How do we hit this when force is true?";
throw new CloudRuntimeException("Caught exception even though it should be handled.", e);
} catch (ConcurrentOperationException e) {
assert false : "How do we hit this when force is true?";
throw new CloudRuntimeException("Caught exception even though it should be handled.", e);
}
}
}
vm = _itMgr.findByIdAndType(vm.getType(), vm.getId());
if (!_forceHA && !vm.isHaEnabled()) {
if (s_logger.isDebugEnabled()) {
s_logger.debug("VM is not HA enabled so we're done.");
}
return null; // VM doesn't require HA
}
if (!_storageMgr.canVmRestartOnAnotherServer(vm.getId())) {
if (s_logger.isDebugEnabled()) {
s_logger.debug("VM can not restart on another server.");
}
return null;
}
if (work.getTimesTried() > _maxRetries) {
s_logger.warn("Retried to max times so deleting: " + vmId);
return null;
}
try {
VMInstanceVO started =
_itMgr.advanceStart(
vm,
new HashMap<VirtualMachineProfile.Param, Object>(),
_accountMgr.getSystemUser(),
_accountMgr.getSystemAccount());
if (started != null) {
s_logger.info("VM is now restarted: " + vmId + " on " + started.getHostId());
return null;
}
if (s_logger.isDebugEnabled()) {
s_logger.debug(
"Rescheduling VM " + vm.toString() + " to try again in " + _restartRetryInterval);
}
} catch (final InsufficientCapacityException e) {
s_logger.warn("Unable to restart " + vm.toString() + " due to " + e.getMessage());
_alertMgr.sendAlert(
alertType,
vm.getDataCenterIdToDeployIn(),
vm.getPodIdToDeployIn(),
"Unable to restart " + vm.getHostName() + " which was running on host " + hostDesc,
"Insufficient capacity to restart VM, name: "
+ vm.getHostName()
+ ", id: "
+ vmId
+ " which was running on host "
+ hostDesc);
} catch (final ResourceUnavailableException e) {
s_logger.warn("Unable to restart " + vm.toString() + " due to " + e.getMessage());
_alertMgr.sendAlert(
alertType,
vm.getDataCenterIdToDeployIn(),
vm.getPodIdToDeployIn(),
"Unable to restart " + vm.getHostName() + " which was running on host " + hostDesc,
"The Storage is unavailable for trying to restart VM, name: "
+ vm.getHostName()
+ ", id: "
+ vmId
+ " which was running on host "
+ hostDesc);
} catch (ConcurrentOperationException e) {
s_logger.warn("Unable to restart " + vm.toString() + " due to " + e.getMessage());
_alertMgr.sendAlert(
alertType,
vm.getDataCenterIdToDeployIn(),
vm.getPodIdToDeployIn(),
"Unable to restart " + vm.getHostName() + " which was running on host " + hostDesc,
"The Storage is unavailable for trying to restart VM, name: "
+ vm.getHostName()
+ ", id: "
+ vmId
+ " which was running on host "
+ hostDesc);
} catch (OperationTimedoutException e) {
s_logger.warn("Unable to restart " + vm.toString() + " due to " + e.getMessage());
_alertMgr.sendAlert(
alertType,
vm.getDataCenterIdToDeployIn(),
vm.getPodIdToDeployIn(),
"Unable to restart " + vm.getHostName() + " which was running on host " + hostDesc,
"The Storage is unavailable for trying to restart VM, name: "
+ vm.getHostName()
+ ", id: "
+ vmId
+ " which was running on host "
+ hostDesc);
}
vm = _itMgr.findByIdAndType(vm.getType(), vm.getId());
work.setUpdateTime(vm.getUpdated());
work.setPreviousState(vm.getState());
return (System.currentTimeMillis() >> 10) + _restartRetryInterval;
}
protected void createDb() {
DataCenterVO dc =
new DataCenterVO(
UUID.randomUUID().toString(),
"test",
"8.8.8.8",
null,
"10.0.0.1",
null,
"10.0.0.1/24",
null,
null,
NetworkType.Basic,
null,
null,
true,
true,
null,
null);
dc = dcDao.persist(dc);
dcId = dc.getId();
HostPodVO pod =
new HostPodVO(UUID.randomUUID().toString(), dc.getId(), "255.255.255.255", "", 8, "test");
pod = podDao.persist(pod);
podId = pod.getId();
ClusterVO cluster = new ClusterVO(dc.getId(), pod.getId(), "devcloud cluster");
cluster.setHypervisorType(HypervisorType.XenServer.toString());
cluster.setClusterType(ClusterType.CloudManaged);
cluster.setManagedState(ManagedState.Managed);
cluster = clusterDao.persist(cluster);
clusterId = cluster.getId();
DataStoreProvider provider =
providerMgr.getDataStoreProvider("ancient primary data store provider");
storage = new StoragePoolVO();
storage.setDataCenterId(dcId);
storage.setPodId(podId);
storage.setPoolType(StoragePoolType.NetworkFilesystem);
storage.setClusterId(clusterId);
storage.setStatus(StoragePoolStatus.Up);
storage.setScope(ScopeType.CLUSTER);
storage.setAvailableBytes(1000);
storage.setCapacityBytes(20000);
storage.setHostAddress(UUID.randomUUID().toString());
storage.setPath(UUID.randomUUID().toString());
storage.setStorageProviderName(provider.getName());
storage = storagePoolDao.persist(storage);
storagePoolId = storage.getId();
storageMgr.createCapacityEntry(storage.getId());
diskOffering = new DiskOfferingVO();
diskOffering.setDiskSize(500);
diskOffering.setName("test-disk");
diskOffering.setSystemUse(false);
diskOffering.setUseLocalStorage(false);
diskOffering.setCustomized(false);
diskOffering.setRecreatable(false);
diskOffering = diskOfferingDao.persist(diskOffering);
diskOfferingId = diskOffering.getId();
volume =
new VolumeVO(
Volume.Type.ROOT,
"volume",
dcId,
1,
1,
diskOffering.getId(),
diskOffering.getDiskSize());
volume = volumeDao.persist(volume);
volumeId = volume.getId();
}
public Long migrate(final HaWorkVO work) {
final long vmId = work.getInstanceId();
final VirtualMachineGuru<VMInstanceVO> mgr = findManager(work.getType());
VMInstanceVO vm = mgr.get(vmId);
if (vm == null || vm.getRemoved() != null) {
s_logger.debug("Unable to find the vm " + vmId);
return null;
}
s_logger.info("Migrating vm: " + vm.toString());
if (vm.getHostId() == null || vm.getHostId() != work.getHostId()) {
s_logger.info("VM is not longer running on the current hostId");
return null;
}
short alertType = AlertManager.ALERT_TYPE_USERVM_MIGRATE;
if (VirtualMachine.Type.DomainRouter.equals(vm.getType())) {
alertType = AlertManager.ALERT_TYPE_DOMAIN_ROUTER_MIGRATE;
} else if (VirtualMachine.Type.ConsoleProxy.equals(vm.getType())) {
alertType = AlertManager.ALERT_TYPE_CONSOLE_PROXY_MIGRATE;
}
HostVO fromHost = _hostDao.findById(vm.getHostId());
String fromHostName = ((fromHost == null) ? "unknown" : fromHost.getName());
HostVO toHost = null;
if (work.getStep() == Step.Scheduled) {
if (vm.getState() != State.Running) {
s_logger.info(
"VM's state is not ready for migration. "
+ vm.toString()
+ " State is "
+ vm.getState().toString());
return (System.currentTimeMillis() >> 10) + _migrateRetryInterval;
}
DataCenterVO dcVO = _dcDao.findById(fromHost.getDataCenterId());
HostPodVO podVO = _podDao.findById(fromHost.getPodId());
try {
toHost = mgr.prepareForMigration(vm);
if (toHost == null) {
if (s_logger.isDebugEnabled()) {
s_logger.debug("Unable to find a host for migrating vm " + vmId);
}
_alertMgr.sendAlert(
alertType,
vm.getDataCenterId(),
vm.getPodId(),
"Unable to migrate vm "
+ vm.getName()
+ " from host "
+ fromHostName
+ " in zone "
+ dcVO.getName()
+ " and pod "
+ podVO.getName(),
"Unable to find a suitable host");
}
} catch (final InsufficientCapacityException e) {
s_logger.warn("Unable to mgirate due to insufficient capacity " + vm.toString());
_alertMgr.sendAlert(
alertType,
vm.getDataCenterId(),
vm.getPodId(),
"Unable to migrate vm "
+ vm.getName()
+ " from host "
+ fromHostName
+ " in zone "
+ dcVO.getName()
+ " and pod "
+ podVO.getName(),
"Insufficient capacity");
} catch (final StorageUnavailableException e) {
s_logger.warn("Storage is unavailable: " + vm.toString());
_alertMgr.sendAlert(
alertType,
vm.getDataCenterId(),
vm.getPodId(),
"Unable to migrate vm "
+ vm.getName()
+ " from host "
+ fromHostName
+ " in zone "
+ dcVO.getName()
+ " and pod "
+ podVO.getName(),
"Storage is gone.");
}
if (toHost == null) {
_agentMgr.maintenanceFailed(vm.getHostId());
return null;
}
if (s_logger.isDebugEnabled()) {
s_logger.debug("Migrating from " + work.getHostId() + " to " + toHost.getId());
}
work.setStep(Step.Migrating);
work.setHostId(toHost.getId());
_haDao.update(work.getId(), work);
}
if (work.getStep() == Step.Migrating) {
vm = mgr.get(vmId); // let's see if anything has changed.
boolean migrated = false;
if (vm == null
|| vm.getRemoved() != null
|| vm.getHostId() == null
|| !_itMgr.stateTransitTo(vm, Event.MigrationRequested, vm.getHostId())) {
s_logger.info("Migration cancelled because state has changed: " + vm.toString());
} else {
try {
boolean isWindows =
_guestOSCategoryDao
.findById(_guestOSDao.findById(vm.getGuestOSId()).getCategoryId())
.getName()
.equalsIgnoreCase("Windows");
MigrateCommand cmd =
new MigrateCommand(vm.getInstanceName(), toHost.getPrivateIpAddress(), isWindows);
Answer answer = _agentMgr.send(fromHost.getId(), cmd);
if (answer != null && answer.getResult()) {
migrated = true;
_storageMgr.unshare(vm, fromHost);
work.setStep(Step.Investigating);
_haDao.update(work.getId(), work);
}
} catch (final AgentUnavailableException e) {
s_logger.debug("host became unavailable");
} catch (final OperationTimedoutException e) {
s_logger.debug("operation timed out");
if (e.isActive()) {
scheduleRestart(vm, true);
}
}
}
if (!migrated) {
s_logger.info("Migration was unsuccessful. Cleaning up: " + vm.toString());
DataCenterVO dcVO = _dcDao.findById(vm.getDataCenterId());
HostPodVO podVO = _podDao.findById(vm.getPodId());
_alertMgr.sendAlert(
alertType,
fromHost.getDataCenterId(),
fromHost.getPodId(),
"Unable to migrate vm "
+ vm.getName()
+ " from host "
+ fromHost.getName()
+ " in zone "
+ dcVO.getName()
+ " and pod "
+ podVO.getName(),
"Migrate Command failed. Please check logs.");
_itMgr.stateTransitTo(vm, Event.MigrationFailedOnSource, toHost.getId());
_agentMgr.maintenanceFailed(vm.getHostId());
Command cleanup = mgr.cleanup(vm, null);
_agentMgr.easySend(toHost.getId(), cleanup);
_storageMgr.unshare(vm, toHost);
return null;
}
}
if (toHost == null) {
toHost = _hostDao.findById(work.getHostId());
}
DataCenterVO dcVO = _dcDao.findById(toHost.getDataCenterId());
HostPodVO podVO = _podDao.findById(toHost.getPodId());
try {
if (!mgr.completeMigration(vm, toHost)) {
_alertMgr.sendAlert(
alertType,
toHost.getDataCenterId(),
toHost.getPodId(),
"Unable to migrate "
+ vmId
+ " to host "
+ toHost.getName()
+ " in zone "
+ dcVO.getName()
+ " and pod "
+ podVO.getName(),
"Migration not completed");
s_logger.warn("Unable to complete migration: " + vm.toString());
} else {
s_logger.info("Migration is complete: " + vm.toString());
}
return null;
} catch (final AgentUnavailableException e) {
s_logger.warn("Agent is unavailable for " + vm.toString());
} catch (final OperationTimedoutException e) {
s_logger.warn("Operation timed outfor " + vm.toString());
}
_itMgr.stateTransitTo(vm, Event.MigrationFailedOnDest, toHost.getId());
return (System.currentTimeMillis() >> 10) + _migrateRetryInterval;
}
protected Long restart(final HaWorkVO work) {
final long vmId = work.getInstanceId();
final VirtualMachineGuru<VMInstanceVO> mgr = findManager(work.getType());
if (mgr == null) {
s_logger.warn(
"Unable to find a handler for " + work.getType().toString() + ", throwing out " + vmId);
return null;
}
VMInstanceVO vm = mgr.get(vmId);
if (vm == null) {
s_logger.info("Unable to find vm: " + vmId);
return null;
}
s_logger.info("HA on " + vm.toString());
if (vm.getState() != work.getPreviousState() || vm.getUpdated() != work.getUpdateTime()) {
s_logger.info(
"VM "
+ vm.toString()
+ " has been changed. Current State = "
+ vm.getState()
+ " Previous State = "
+ work.getPreviousState()
+ " last updated = "
+ vm.getUpdated()
+ " previous updated = "
+ work.getUpdateTime());
return null;
}
final HostVO host = _hostDao.findById(work.getHostId());
DataCenterVO dcVO = _dcDao.findById(host.getDataCenterId());
HostPodVO podVO = _podDao.findById(host.getPodId());
String hostDesc =
"name: "
+ host.getName()
+ "(id:"
+ host.getId()
+ "), availability zone: "
+ dcVO.getName()
+ ", pod: "
+ podVO.getName();
short alertType = AlertManager.ALERT_TYPE_USERVM;
if (VirtualMachine.Type.DomainRouter.equals(vm.getType())) {
alertType = AlertManager.ALERT_TYPE_DOMAIN_ROUTER;
} else if (VirtualMachine.Type.ConsoleProxy.equals(vm.getType())) {
alertType = AlertManager.ALERT_TYPE_CONSOLE_PROXY;
}
Boolean alive = null;
if (work.getStep() == Step.Investigating) {
if (vm.getHostId() == null || vm.getHostId() != work.getHostId()) {
s_logger.info("VM " + vm.toString() + " is now no longer on host " + work.getHostId());
if (vm.getState() == State.Starting && vm.getUpdated() == work.getUpdateTime()) {
_itMgr.stateTransitTo(vm, Event.AgentReportStopped, null);
}
return null;
}
Enumeration<Investigator> en = _investigators.enumeration();
Investigator investigator = null;
while (en.hasMoreElements()) {
investigator = en.nextElement();
alive = investigator.isVmAlive(vm, host);
if (alive != null) {
s_logger.debug(
investigator.getName() + " found VM " + vm.getName() + "to be alive? " + alive);
break;
}
}
if (alive != null && alive) {
s_logger.debug("VM " + vm.getName() + " is found to be alive by " + investigator.getName());
if (host.getStatus() == Status.Up) {
compareState(vm, new AgentVmInfo(vm.getInstanceName(), mgr, State.Running), false);
return null;
} else {
s_logger.debug("Rescheduling because the host is not up but the vm is alive");
return (System.currentTimeMillis() >> 10) + _investigateRetryInterval;
}
}
boolean fenced = false;
if (alive == null || !alive) {
fenced = true;
s_logger.debug("Fencing off VM that we don't know the state of");
Enumeration<FenceBuilder> enfb = _fenceBuilders.enumeration();
while (enfb.hasMoreElements()) {
final FenceBuilder fb = enfb.nextElement();
Boolean result = fb.fenceOff(vm, host);
if (result != null && !result) {
fenced = false;
}
}
}
if (alive == null && !fenced) {
s_logger.debug("We were unable to fence off the VM " + vm.toString());
_alertMgr.sendAlert(
alertType,
vm.getDataCenterId(),
vm.getPodId(),
"Unable to restart " + vm.getName() + " which was running on host " + hostDesc,
"Insufficient capacity to restart VM, name: "
+ vm.getName()
+ ", id: "
+ vmId
+ " which was running on host "
+ hostDesc);
return (System.currentTimeMillis() >> 10) + _restartRetryInterval;
}
mgr.completeStopCommand(vm);
work.setStep(Step.Scheduled);
_haDao.update(work.getId(), work);
}
// send an alert for VMs that stop unexpectedly
_alertMgr.sendAlert(
alertType,
vm.getDataCenterId(),
vm.getPodId(),
"VM (name: "
+ vm.getName()
+ ", id: "
+ vmId
+ ") stopped unexpectedly on host "
+ hostDesc,
"Virtual Machine "
+ vm.getName()
+ " (id: "
+ vm.getId()
+ ") running on host ["
+ hostDesc
+ "] stopped unexpectedly.");
vm = mgr.get(vm.getId());
if (!_forceHA && !vm.isHaEnabled()) {
if (s_logger.isDebugEnabled()) {
s_logger.debug("VM is not HA enabled so we're done.");
}
return null; // VM doesn't require HA
}
if (!_storageMgr.canVmRestartOnAnotherServer(vm.getId())) {
if (s_logger.isDebugEnabled()) {
s_logger.debug("VM can not restart on another server.");
}
return null;
}
if (work.getTimesTried() > _maxRetries) {
s_logger.warn("Retried to max times so deleting: " + vmId);
return null;
}
try {
VMInstanceVO started = mgr.start(vm.getId(), 0);
if (started != null) {
s_logger.info("VM is now restarted: " + vmId + " on " + started.getHostId());
return null;
}
if (s_logger.isDebugEnabled()) {
s_logger.debug(
"Rescheduling VM " + vm.toString() + " to try again in " + _restartRetryInterval);
}
vm = mgr.get(vm.getId());
work.setUpdateTime(vm.getUpdated());
work.setPreviousState(vm.getState());
return (System.currentTimeMillis() >> 10) + _restartRetryInterval;
} catch (final InsufficientCapacityException e) {
s_logger.warn("Unable to restart " + vm.toString() + " due to " + e.getMessage());
_alertMgr.sendAlert(
alertType,
vm.getDataCenterId(),
vm.getPodId(),
"Unable to restart " + vm.getName() + " which was running on host " + hostDesc,
"Insufficient capacity to restart VM, name: "
+ vm.getName()
+ ", id: "
+ vmId
+ " which was running on host "
+ hostDesc);
return null;
} catch (final StorageUnavailableException e) {
s_logger.warn("Unable to restart " + vm.toString() + " due to " + e.getMessage());
_alertMgr.sendAlert(
alertType,
vm.getDataCenterId(),
vm.getPodId(),
"Unable to restart " + vm.getName() + " which was running on host " + hostDesc,
"The Storage is unavailable for trying to restart VM, name: "
+ vm.getName()
+ ", id: "
+ vmId
+ " which was running on host "
+ hostDesc);
return null;
} catch (ConcurrentOperationException e) {
s_logger.warn("Unable to restart " + vm.toString() + " due to " + e.getMessage());
_alertMgr.sendAlert(
alertType,
vm.getDataCenterId(),
vm.getPodId(),
"Unable to restart " + vm.getName() + " which was running on host " + hostDesc,
"The Storage is unavailable for trying to restart VM, name: "
+ vm.getName()
+ ", id: "
+ vmId
+ " which was running on host "
+ hostDesc);
return null;
} catch (ExecutionException e) {
s_logger.warn("Unable to restart " + vm.toString() + " due to " + e.getMessage());
_alertMgr.sendAlert(
alertType,
vm.getDataCenterId(),
vm.getPodId(),
"Unable to restart " + vm.getName() + " which was running on host " + hostDesc,
"The Storage is unavailable for trying to restart VM, name: "
+ vm.getName()
+ ", id: "
+ vmId
+ " which was running on host "
+ hostDesc);
return null;
}
}
@Override
@DB
public void recalculateCapacity() {
// FIXME: the right way to do this is to register a listener (see RouterStatsListener,
// VMSyncListener)
// for the vm sync state. The listener model has connects/disconnects to keep things in
// sync much better
// than this model right now, so when a VM is started, we update the amount allocated,
// and when a VM
// is stopped we updated the amount allocated, and when VM sync reports a changed state,
// we update
// the amount allocated. Hopefully it's limited to 3 entry points and will keep the
// amount allocated
// per host accurate.
try {
if (s_logger.isDebugEnabled()) {
s_logger.debug("recalculating system capacity");
s_logger.debug("Executing cpu/ram capacity update");
}
// Calculate CPU and RAM capacities
// get all hosts...even if they are not in 'UP' state
List<HostVO> hosts = _resourceMgr.listAllHostsInAllZonesByType(Host.Type.Routing);
for (HostVO host : hosts) {
_capacityMgr.updateCapacityForHost(host);
}
if (s_logger.isDebugEnabled()) {
s_logger.debug("Done executing cpu/ram capacity update");
s_logger.debug("Executing storage capacity update");
}
// Calculate storage pool capacity
List<StoragePoolVO> storagePools = _storagePoolDao.listAll();
for (StoragePoolVO pool : storagePools) {
long disk = 0l;
Pair<Long, Long> sizes = _volumeDao.getCountAndTotalByPool(pool.getId());
disk = sizes.second();
if (pool.isShared()) {
_storageMgr.createCapacityEntry(pool, Capacity.CAPACITY_TYPE_STORAGE_ALLOCATED, disk);
} else {
_storageMgr.createCapacityEntry(pool, Capacity.CAPACITY_TYPE_LOCAL_STORAGE, disk);
}
}
if (s_logger.isDebugEnabled()) {
s_logger.debug("Done executing storage capacity update");
s_logger.debug("Executing capacity updates public ip and Vlans");
}
List<DataCenterVO> datacenters = _dcDao.listAll();
for (DataCenterVO datacenter : datacenters) {
long dcId = datacenter.getId();
// NOTE
// What happens if we have multiple vlans? Dashboard currently shows stats
// with no filter based on a vlan
// ideal way would be to remove out the vlan param, and filter only on dcId
// implementing the same
// Calculate new Public IP capacity for Virtual Network
if (datacenter.getNetworkType() == NetworkType.Advanced) {
createOrUpdateIpCapacity(dcId, null, CapacityVO.CAPACITY_TYPE_VIRTUAL_NETWORK_PUBLIC_IP);
}
// Calculate new Public IP capacity for Direct Attached Network
createOrUpdateIpCapacity(dcId, null, CapacityVO.CAPACITY_TYPE_DIRECT_ATTACHED_PUBLIC_IP);
if (datacenter.getNetworkType() == NetworkType.Advanced) {
// Calculate VLAN's capacity
createOrUpdateVlanCapacity(dcId);
}
}
if (s_logger.isDebugEnabled()) {
s_logger.debug("Done capacity updates for public ip and Vlans");
s_logger.debug("Executing capacity updates for private ip");
}
// Calculate new Private IP capacity
List<HostPodVO> pods = _podDao.listAll();
for (HostPodVO pod : pods) {
long podId = pod.getId();
long dcId = pod.getDataCenterId();
createOrUpdateIpCapacity(dcId, podId, CapacityVO.CAPACITY_TYPE_PRIVATE_IP);
}
if (s_logger.isDebugEnabled()) {
s_logger.debug("Done executing capacity updates for private ip");
s_logger.debug("Done recalculating system capacity");
}
} catch (Throwable t) {
s_logger.error("Caught exception in recalculating capacity", t);
}
}