@Override
public void scheduleStop(VMInstanceVO vm, long hostId, WorkType type) {
assert (type == WorkType.CheckStop || type == WorkType.ForceStop || type == WorkType.Stop);
if (_haDao.hasBeenScheduled(vm.getId(), type)) {
s_logger.info("There's already a job scheduled to stop " + vm);
return;
}
HaWorkVO work =
new HaWorkVO(
vm.getId(),
vm.getType(),
type,
Step.Scheduled,
hostId,
vm.getState(),
0,
vm.getUpdated());
_haDao.persist(work);
if (s_logger.isDebugEnabled()) {
s_logger.debug("Scheduled " + work);
}
wakeupWorkers();
}
public Long migrate(final HaWorkVO work) {
long vmId = work.getInstanceId();
long srcHostId = work.getHostId();
try {
work.setStep(Step.Migrating);
_haDao.update(work.getId(), work);
VMInstanceVO vm = _instanceDao.findById(vmId);
// First try starting the vm with its original planner, if it doesn't succeed send HAPlanner
// as its an emergency.
boolean result = false;
try {
_itMgr.migrateAway(vm.getUuid(), srcHostId, null);
} catch (InsufficientServerCapacityException e) {
s_logger.warn("Failed to deploy vm " + vmId + " with original planner, sending HAPlanner");
_itMgr.migrateAway(vm.getUuid(), srcHostId, _haPlanners.get(0));
}
return null;
} catch (InsufficientServerCapacityException e) {
s_logger.warn("Insufficient capacity for migrating a VM.");
_resourceMgr.maintenanceFailed(srcHostId);
return (System.currentTimeMillis() >> 10) + _migrateRetryInterval;
}
}
@Override
public List<VMInstanceVO> findTakenMigrationWork() {
List<HaWorkVO> works = _haDao.findTakenWorkItems(WorkType.Migration);
List<VMInstanceVO> vms = new ArrayList<VMInstanceVO>(works.size());
for (HaWorkVO work : works) {
vms.add(_instanceDao.findById(work.getInstanceId()));
}
return vms;
}
@Override
public boolean configure(final String name, final Map<String, Object> xmlParams)
throws ConfigurationException {
_serverId = _msServer.getId();
Map<String, String> params = new HashMap<String, String>();
params = _configDao.getConfiguration(Long.toHexString(_serverId), xmlParams);
String value = params.get(Config.HAWorkers.key());
final int count = NumbersUtil.parseInt(value, 1);
_workers = new WorkerThread[count];
for (int i = 0; i < _workers.length; i++) {
_workers[i] = new WorkerThread("HA-Worker-" + i);
}
value = params.get("force.ha");
_forceHA = Boolean.parseBoolean(value);
value = params.get("time.to.sleep");
_timeToSleep = NumbersUtil.parseInt(value, 60) * 1000;
value = params.get("max.retries");
_maxRetries = NumbersUtil.parseInt(value, 5);
value = params.get("time.between.failures");
_timeBetweenFailures = NumbersUtil.parseLong(value, 3600) * 1000;
value = params.get("time.between.cleanup");
_timeBetweenCleanups = NumbersUtil.parseLong(value, 3600 * 24);
value = params.get("stop.retry.interval");
_stopRetryInterval = NumbersUtil.parseInt(value, 10 * 60);
value = params.get("restart.retry.interval");
_restartRetryInterval = NumbersUtil.parseInt(value, 10 * 60);
value = params.get("investigate.retry.interval");
_investigateRetryInterval = NumbersUtil.parseInt(value, 1 * 60);
value = params.get("migrate.retry.interval");
_migrateRetryInterval = NumbersUtil.parseInt(value, 2 * 60);
_instance = params.get("instance");
if (_instance == null) {
_instance = "VMOPS";
}
_haTag = params.get("ha.tag");
_haDao.releaseWorkItems(_serverId);
_stopped = true;
_executor = Executors.newScheduledThreadPool(count, new NamedThreadFactory("HA"));
return true;
}
@Override
public void scheduleStop(final VMInstanceVO vm, long hostId, boolean verifyHost) {
if (_haDao.hasBeenScheduled(vm.getId(), verifyHost ? WorkType.CheckStop : WorkType.Stop)) {
s_logger.info("There's already a job scheduled to stop " + vm.toString());
return;
}
final HaWorkVO work =
new HaWorkVO(
vm.getId(),
vm.getType(),
verifyHost ? WorkType.CheckStop : WorkType.Stop,
Step.Scheduled,
hostId,
vm.getState(),
0,
vm.getUpdated());
_haDao.persist(work);
if (s_logger.isDebugEnabled()) {
s_logger.debug("Scheduled " + work.toString() + " verifyHost = " + verifyHost);
}
wakeupWorkers();
}
@Override
public boolean scheduleMigration(final VMInstanceVO vm) {
final HaWorkVO work =
new HaWorkVO(
vm.getId(),
vm.getType(),
WorkType.Migration,
Step.Scheduled,
vm.getHostId(),
vm.getState(),
0,
vm.getUpdated());
_haDao.persist(work);
wakeupWorkers();
return true;
}
@Override
public void scheduleDestroy(VMInstanceVO vm, long hostId) {
final HaWorkVO work =
new HaWorkVO(
vm.getId(),
vm.getType(),
WorkType.Destroy,
Step.Scheduled,
hostId,
vm.getState(),
0,
vm.getUpdated());
_haDao.persist(work);
if (s_logger.isDebugEnabled()) {
s_logger.debug("Scheduled " + work.toString());
}
wakeupWorkers();
}
public Long migrate(final HaWorkVO work) {
long vmId = work.getInstanceId();
long srcHostId = work.getHostId();
try {
work.setStep(Step.Migrating);
_haDao.update(work.getId(), work);
if (!_itMgr.migrateAway(work.getType(), vmId, srcHostId)) {
s_logger.warn("Unable to migrate vm from " + srcHostId);
_resourceMgr.maintenanceFailed(srcHostId);
}
return null;
} catch (InsufficientServerCapacityException e) {
s_logger.warn("Insufficient capacity for migrating a VM.");
_resourceMgr.maintenanceFailed(srcHostId);
return (System.currentTimeMillis() >> 10) + _migrateRetryInterval;
} catch (VirtualMachineMigrationException e) {
s_logger.warn("Looks like VM is still starting, we need to retry migrating the VM later.");
_resourceMgr.maintenanceFailed(srcHostId);
return (System.currentTimeMillis() >> 10) + _migrateRetryInterval;
}
}
@Override
public void cancelScheduledMigrations(final HostVO host) {
WorkType type = host.getType() == HostVO.Type.Storage ? WorkType.Stop : WorkType.Migration;
_haDao.deleteMigrationWorkItems(host.getId(), type, _serverId);
}
@Override
public void cancelDestroy(VMInstanceVO vm, Long hostId) {
_haDao.delete(vm.getId(), WorkType.Destroy);
}
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;
}
@Override
public void scheduleRestart(VMInstanceVO vm, boolean investigate) {
Long hostId = vm.getHostId();
if (hostId == null) {
try {
s_logger.debug("Found a vm that is scheduled to be restarted but has no host id: " + vm);
_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);
}
return;
}
if (vm.getHypervisorType() == HypervisorType.VMware) {
s_logger.info("Skip HA for VMware VM " + vm.getInstanceName());
return;
}
if (!investigate) {
if (s_logger.isDebugEnabled()) {
s_logger.debug(
"VM does not require investigation so I'm marking it as Stopped: " + vm.toString());
}
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;
}
if (!(_forceHA || vm.isHaEnabled())) {
String hostDesc =
"id:"
+ vm.getHostId()
+ ", availability zone id:"
+ vm.getDataCenterIdToDeployIn()
+ ", pod id:"
+ vm.getPodIdToDeployIn();
_alertMgr.sendAlert(
alertType,
vm.getDataCenterIdToDeployIn(),
vm.getPodIdToDeployIn(),
"VM (name: "
+ vm.getHostName()
+ ", id: "
+ vm.getId()
+ ") stopped unexpectedly on host "
+ hostDesc,
"Virtual Machine "
+ vm.getHostName()
+ " (id: "
+ vm.getId()
+ ") running on host ["
+ vm.getHostId()
+ "] stopped unexpectedly.");
if (s_logger.isDebugEnabled()) {
s_logger.debug("VM is not HA enabled so we're done.");
}
}
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);
}
}
List<HaWorkVO> items = _haDao.findPreviousHA(vm.getId());
int maxRetries = 0;
for (HaWorkVO item : items) {
if (maxRetries < item.getTimesTried() && !item.canScheduleNew(_timeBetweenFailures)) {
maxRetries = item.getTimesTried();
break;
}
}
HaWorkVO work =
new HaWorkVO(
vm.getId(),
vm.getType(),
WorkType.HA,
investigate ? Step.Investigating : Step.Scheduled,
hostId,
vm.getState(),
maxRetries + 1,
vm.getUpdated());
_haDao.persist(work);
if (s_logger.isInfoEnabled()) {
s_logger.info("Schedule vm for HA: " + vm);
}
wakeupWorkers();
}
@Override
public void onManagementNodeLeft(List<ManagementServerHostVO> nodeList, long selfNodeId) {
for (ManagementServerHostVO node : nodeList) {
_haDao.releaseWorkItems(node.getMsid());
}
}
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
public void scheduleRestart(VMInstanceVO vm, final boolean investigate) {
Long hostId = vm.getHostId();
VirtualMachineGuru<VMInstanceVO> mgr = findManager(vm.getType());
vm = mgr.get(vm.getId());
if (!investigate) {
if (s_logger.isDebugEnabled()) {
s_logger.debug(
"VM does not require investigation so I'm marking it as Stopped: " + vm.toString());
}
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;
}
if (!(_forceHA || vm.isHaEnabled())) {
_alertMgr.sendAlert(
alertType,
vm.getDataCenterId(),
vm.getPodId(),
"VM (name: "
+ vm.getName()
+ ", id: "
+ vm.getId()
+ ") stopped unexpectedly on host "
+ vm.getHostId(),
"Virtual Machine "
+ vm.getName()
+ " (id: "
+ vm.getId()
+ ") running on host ["
+ vm.getHostId()
+ "] stopped unexpectedly.");
if (s_logger.isDebugEnabled()) {
s_logger.debug("VM is not HA enabled so we're done.");
}
}
mgr.completeStopCommand(vm);
}
final List<HaWorkVO> items = _haDao.findPreviousHA(vm.getId());
int maxRetries = 0;
for (final HaWorkVO item : items) {
if (maxRetries < item.getTimesTried() && !item.canScheduleNew(_timeBetweenFailures)) {
maxRetries = item.getTimesTried();
break;
}
}
final HaWorkVO work =
new HaWorkVO(
vm.getId(),
vm.getType(),
WorkType.HA,
investigate ? Step.Investigating : Step.Scheduled,
hostId,
vm.getState(),
maxRetries + 1,
vm.getUpdated());
_haDao.persist(work);
if (s_logger.isInfoEnabled()) {
s_logger.info("Schedule vm for HA: " + vm.toString());
}
wakeupWorkers();
}
