Skip to content

Linstor fix host picking #12047

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 2 commits into
base: 4.20
Choose a base branch
from
Open

Linstor fix host picking #12047

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
425d257
linstor: improve aborting copyAsync
rp- Nov 7, 2025
1e9040f
linstor: Pick correct host for agent operations
rp- Nov 7, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
98 changes: 52 additions & 46 deletions ...ava/org/apache/cloudstack/storage/datastore/driver/LinstorPrimaryDataStoreDriverImpl.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -63,6 +63,7 @@
import com.cloud.api.storage.LinstorRevertBackupSnapshotCommand;
import com.cloud.configuration.Config;
import com.cloud.host.Host;
import com.cloud.host.HostVO;
import com.cloud.host.dao.HostDao;
import com.cloud.resource.ResourceState;
import com.cloud.storage.DataStoreRole;
Expand Down Expand Up @@ -921,9 +922,10 @@ private String revertSnapshotFromImageStore(
_backupsnapshotwait,
VirtualMachineManager.ExecuteInSequence.value());

Optional<RemoteHostEndPoint> optEP = getDiskfullEP(linstorApi, rscName);
final StoragePool pool = (StoragePool) volumeInfo.getDataStore();
Optional<RemoteHostEndPoint> optEP = getDiskfullEP(linstorApi, pool, rscName);
if (optEP.isEmpty()) {
optEP = getLinstorEP(linstorApi, rscName);
optEP = getLinstorEP(linstorApi, pool, rscName);
}

if (optEP.isPresent()) {
Expand Down Expand Up @@ -1063,13 +1065,27 @@ public void copyAsync(DataObject srcData, DataObject dstData, AsyncCompletionCal
Answer answer = copyVolume(srcData, dstData);
res = new CopyCommandResult(null, answer);
} else {
Answer answer = new Answer(null, false, "noimpl");
res = new CopyCommandResult(null, answer);
res.setResult("Not implemented yet");
throw new CloudRuntimeException("Not implemented for Linstor primary storage.");
}
callback.complete(res);
}

private Host getEnabledClusterHost(StoragePool storagePool, List<String> linstorNodeNames) {
List<HostVO> csHosts;
if (storagePool.getClusterId() != null) {
csHosts = _hostDao.findByClusterId(storagePool.getClusterId());
} else {
csHosts = _hostDao.findByDataCenterId(storagePool.getDataCenterId());
}
Collections.shuffle(csHosts); // so we do not always pick the same host for operations
for (HostVO host : csHosts) {
if (host.getResourceState() == ResourceState.Enabled && linstorNodeNames.contains(host.getName())) {
return host;
}
}
return null;
}

/**
* Tries to get a Linstor cloudstack end point, that is at least diskless.
*
Expand All @@ -1078,47 +1094,37 @@ public void copyAsync(DataObject srcData, DataObject dstData, AsyncCompletionCal
* @return Optional RemoteHostEndPoint if one could get found.
* @throws ApiException
*/
private Optional<RemoteHostEndPoint> getLinstorEP(DevelopersApi api, String rscName) throws ApiException {
private Optional<RemoteHostEndPoint> getLinstorEP(DevelopersApi api, StoragePool storagePool, String rscName)
throws ApiException {
List<String> linstorNodeNames = LinstorUtil.getLinstorNodeNames(api);
Collections.shuffle(linstorNodeNames); // do not always pick the first linstor node

Host host = null;
for (String nodeName : linstorNodeNames) {
host = _hostDao.findByName(nodeName);
if (host != null && host.getResourceState() == ResourceState.Enabled) {
logger.info(String.format("Linstor: Make resource %s available on node %s ...", rscName, nodeName));
ApiCallRcList answers = api.resourceMakeAvailableOnNode(rscName, nodeName, new ResourceMakeAvailable());
if (!answers.hasError()) {
break; // found working host
} else {
logger.error(
String.format("Linstor: Unable to make resource %s on node %s available: %s",
rscName,
nodeName,
LinstorUtil.getBestErrorMessage(answers)));
}
Host host = getEnabledClusterHost(storagePool, linstorNodeNames);
if (host != null) {
logger.info("Linstor: Make resource {} available on node {} ...", rscName, host.getName());
ApiCallRcList answers = api.resourceMakeAvailableOnNode(
rscName, host.getName(), new ResourceMakeAvailable());
if (answers.hasError()) {
logger.error("Linstor: Unable to make resource {} on node {} available: {}",
rscName, host.getName(), LinstorUtil.getBestErrorMessage(answers));
return Optional.empty();
} else {
return Optional.of(RemoteHostEndPoint.getHypervisorHostEndPoint(host));
}
}

if (host == null)
{
logger.error("Linstor: Couldn't create a resource on any cloudstack host.");
return Optional.empty();
}
else
{
return Optional.of(RemoteHostEndPoint.getHypervisorHostEndPoint(host));
}
logger.error("Linstor: Couldn't create a resource on any cloudstack host.");
return Optional.empty();
}

private Optional<RemoteHostEndPoint> getDiskfullEP(DevelopersApi api, String rscName) throws ApiException {
private Optional<RemoteHostEndPoint> getDiskfullEP(DevelopersApi api, StoragePool storagePool, String rscName)
throws ApiException {
List<com.linbit.linstor.api.model.StoragePool> linSPs = LinstorUtil.getDiskfulStoragePools(api, rscName);
if (linSPs != null) {
for (com.linbit.linstor.api.model.StoragePool sp : linSPs) {
Host host = _hostDao.findByName(sp.getNodeName());
if (host != null && host.getResourceState() == ResourceState.Enabled) {
return Optional.of(RemoteHostEndPoint.getHypervisorHostEndPoint(host));
}
List<String> linstorNodeNames = linSPs.stream()
.map(com.linbit.linstor.api.model.StoragePool::getNodeName)
.collect(Collectors.toList());
Host host = getEnabledClusterHost(storagePool, linstorNodeNames);
if (host != null) {
return Optional.of(RemoteHostEndPoint.getHypervisorHostEndPoint(host));
}
}
logger.error("Linstor: No diskfull host found.");
Expand Down Expand Up @@ -1199,12 +1205,12 @@ private Answer copyTemplate(DataObject srcData, DataObject dstData) {
VirtualMachineManager.ExecuteInSequence.value());

try {
Optional<RemoteHostEndPoint> optEP = getLinstorEP(api, rscName);
Optional<RemoteHostEndPoint> optEP = getLinstorEP(api, pool, rscName);
if (optEP.isPresent()) {
answer = optEP.get().sendMessage(cmd);
} else {
answer = new Answer(cmd, false, "Unable to get matching Linstor endpoint.");
deleteResourceDefinition(pool, rscName);
throw new CloudRuntimeException("Unable to get matching Linstor endpoint.");
}
} catch (ApiException exc) {
logger.error("copy template failed: ", exc);
Expand Down Expand Up @@ -1241,12 +1247,12 @@ private Answer copyVolume(DataObject srcData, DataObject dstData) {
Answer answer;

try {
Optional<RemoteHostEndPoint> optEP = getLinstorEP(api, rscName);
Optional<RemoteHostEndPoint> optEP = getLinstorEP(api, pool, rscName);
if (optEP.isPresent()) {
answer = optEP.get().sendMessage(cmd);
}
else {
answer = new Answer(cmd, false, "Unable to get matching Linstor endpoint.");
throw new CloudRuntimeException("Unable to get matching Linstor endpoint.");
}
} catch (ApiException exc) {
logger.error("copy volume failed: ", exc);
Expand Down Expand Up @@ -1279,14 +1285,14 @@ private Answer copyFromTemporaryResource(
try {
String devName = restoreResourceFromSnapshot(api, pool, rscName, snapshotName, restoreName);

Optional<RemoteHostEndPoint> optEPAny = getLinstorEP(api, restoreName);
Optional<RemoteHostEndPoint> optEPAny = getLinstorEP(api, pool, restoreName);
if (optEPAny.isPresent()) {
// patch the src device path to the temporary linstor resource
snapshotObject.setPath(devName);
origCmd.setSrcTO(snapshotObject.getTO());
answer = optEPAny.get().sendMessage(origCmd);
} else{
answer = new Answer(origCmd, false, "Unable to get matching Linstor endpoint.");
} else {
throw new CloudRuntimeException("Unable to get matching Linstor endpoint.");
}
} finally {
// delete the temporary resource, noop if already gone
Expand Down Expand Up @@ -1348,7 +1354,7 @@ protected Answer copySnapshot(DataObject srcData, DataObject destData) {
VirtualMachineManager.ExecuteInSequence.value());
cmd.setOptions(options);

Optional<RemoteHostEndPoint> optEP = getDiskfullEP(api, rscName);
Optional<RemoteHostEndPoint> optEP = getDiskfullEP(api, pool, rscName);
Answer answer;
if (optEP.isPresent()) {
answer = optEP.get().sendMessage(cmd);
Expand Down