Skip to content
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

Jump to bottom

OWLS-89106 - Potential fix for pod startup issue in GBU CNE environment after node drain/repave operation #2398

Merged
merged 23 commits into from
Jun 16, 2021
Merged

OWLS-89106 - Potential fix for pod startup issue in GBU CNE environment after node drain/repave operation #2398

Show file tree
Hide file tree
Changes from 7 commits
Commits
Show all changes
23 commits
Select commit Hold shift + click to select a range
b33a4a3
Potential fix for pod startup issue in GBU CNE env after node drain/r…
ankedia Jun 8, 2021
9c83cb0
Increase the pod ready wait timeout to 5 min (instead of 2 min) to av…
ankedia Jun 9, 2021
4d6749c
Undo refactoring related changes.
ankedia Jun 9, 2021
291d6d2
Fix to not increment count for introspector job.
ankedia Jun 10, 2021
6d106c7
Potential fix for the race condition in the introspector job completi…
ankedia Jun 10, 2021
3fb128f
Reduce timeout interval to 2 min.
ankedia Jun 10, 2021
70eab49
Add previously removed event validation.
ankedia Jun 11, 2021
4dbbe1c
Changes to check if the cached pod is not found on read then execute …
ankedia Jun 12, 2021
240130d
Fix checkstyle error.
ankedia Jun 12, 2021
b035257
Minor refactoring and increase the timeout value.
ankedia Jun 12, 2021
e988050
Fix comments and minor change.
ankedia Jun 12, 2021
b009898
Minor changes.
ankedia Jun 14, 2021
9a913de
Fix checkstyle error.
ankedia Jun 14, 2021
2093c8f
PR review comments - create new MakeRightDomainOperation instead of u…
ankedia Jun 14, 2021
b3f2362
Revert previous change to create a new make-right operation and imple…
ankedia Jun 14, 2021
b471058
Refactoring changes.
ankedia Jun 14, 2021
7b2bc9f
Check for null call result.
ankedia Jun 15, 2021
5a040cc
More refactoring changes.
ankedia Jun 15, 2021
b974ff8
Added debug message.
ankedia Jun 15, 2021
0616eb6
Minor changes.
ankedia Jun 15, 2021
1c3b6a0
added unit test to test that wait for ready timeout executes make rig…
ankedia Jun 16, 2021
a61ddea
Minor change.
ankedia Jun 16, 2021
23b60ae
Remove unused variable and potential fix for integration test failure.
ankedia Jun 16, 2021
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
8 changes: 7 additions & 1 deletion operator/src/main/java/oracle/kubernetes/operator/DomainProcessorImpl.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -746,7 +746,7 @@ private void addServerToMaps(Map<String, ServerHealth> serverHealthMap,
*/
class MakeRightDomainOperationImpl implements MakeRightDomainOperation {

private final DomainPresenceInfo liveInfo;
private DomainPresenceInfo liveInfo;
private boolean explicitRecheck;
private boolean deleting;
private boolean willInterrupt;
Expand Down Expand Up @@ -851,6 +851,12 @@ public void setInspectionRun() {
inspectionRun = true;
}

@Override
public void setLiveInfo(DomainPresenceInfo info) {
this.liveInfo = info;
}


@Override
public boolean wasInspectionRun() {
return inspectionRun;
Expand Down
2 changes: 2 additions & 0 deletions operator/src/main/java/oracle/kubernetes/operator/MakeRightDomainOperation.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,6 +36,8 @@ public interface MakeRightDomainOperation {

void setInspectionRun();

void setLiveInfo(DomainPresenceInfo info);

boolean wasInspectionRun();

private static boolean wasInspectionRun(Packet packet) {
Expand Down
5 changes: 4 additions & 1 deletion operator/src/main/java/oracle/kubernetes/operator/TuningParameters.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -194,17 +194,20 @@ class WatchTuning {
public final int watchLifetime;
public final int watchMinimumDelay;
public final int watchBackstopRecheckDelay;
public final int watchBackstopRecheckCount;

/**
* Create watch tuning.
* @param watchLifetime Watch lifetime
* @param watchMinimumDelay Minimum delay before accepting new events to prevent hot loops
* @param watchBackstopRecheckDelay Recheck delay for get while waiting for a status to backstop missed watch events
*/
public WatchTuning(int watchLifetime, int watchMinimumDelay, int watchBackstopRecheckDelay) {
public WatchTuning(int watchLifetime, int watchMinimumDelay, int watchBackstopRecheckDelay,
int watchBackstopRecheckCount) {
this.watchLifetime = watchLifetime;
this.watchMinimumDelay = watchMinimumDelay;
this.watchBackstopRecheckDelay = watchBackstopRecheckDelay;
this.watchBackstopRecheckCount = watchBackstopRecheckCount;
}

@Override
Expand Down
3 changes: 2 additions & 1 deletion operator/src/main/java/oracle/kubernetes/operator/TuningParametersImpl.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -75,7 +75,8 @@ private void update() {
new WatchTuning(
(int) readTuningParameter("watchLifetime", 300),
(int) readTuningParameter("watchMinimumDelay", 5),
(int) readTuningParameter("watchBackstopRecheckDelaySeconds", 5));
(int) readTuningParameter("watchBackstopRecheckDelaySeconds", 5),
(int) readTuningParameter("watchBackstopRecheckCount", 24));

PodTuning pod =
new PodTuning(
Expand Down
54 changes: 50 additions & 4 deletions operator/src/main/java/oracle/kubernetes/operator/WaitForReadyStep.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,17 +8,21 @@
import java.util.concurrent.atomic.AtomicBoolean;
import java.util.function.Consumer;

import io.kubernetes.client.openapi.models.V1Job;
import io.kubernetes.client.openapi.models.V1ObjectMeta;
import io.kubernetes.client.openapi.models.V1Pod;
import oracle.kubernetes.operator.calls.CallResponse;
import oracle.kubernetes.operator.helpers.CallBuilder;
import oracle.kubernetes.operator.helpers.DomainPresenceInfo;
import oracle.kubernetes.operator.helpers.ResponseStep;
import oracle.kubernetes.operator.steps.DefaultResponseStep;
import oracle.kubernetes.operator.work.AsyncFiber;
import oracle.kubernetes.operator.work.NextAction;
import oracle.kubernetes.operator.work.Packet;
import oracle.kubernetes.operator.work.Step;
import oracle.kubernetes.weblogic.domain.model.Domain;

import static oracle.kubernetes.operator.ProcessingConstants.MAKE_RIGHT_DOMAIN_OPERATION;
import static oracle.kubernetes.operator.helpers.KubernetesUtils.getDomainUidLabel;

/**
Expand All @@ -28,6 +32,7 @@
*/
abstract class WaitForReadyStep<T> extends Step {
private static final int DEFAULT_RECHECK_SECONDS = 5;
private static final int DEFAULT_RECHECK_COUNT = 24;

static int getWatchBackstopRecheckDelaySeconds() {
return Optional.ofNullable(TuningParameters.getInstance())
Expand All @@ -36,6 +41,13 @@ static int getWatchBackstopRecheckDelaySeconds() {
.orElse(DEFAULT_RECHECK_SECONDS);
}

static int getWatchBackstopRecheckCount() {
return Optional.ofNullable(TuningParameters.getInstance())
.map(TuningParameters::getWatchTuning)
.map(t -> t.watchBackstopRecheckCount)
.orElse(DEFAULT_RECHECK_COUNT);
}

private final T initialResource;
private final String resourceName;

Expand Down Expand Up @@ -197,17 +209,32 @@ private DefaultResponseStep<T> resumeIfReady(Callback callback) {
return new DefaultResponseStep<>(null) {
@Override
public NextAction onSuccess(Packet packet, CallResponse<T> callResponse) {
DomainPresenceInfo info = packet.getSpi(DomainPresenceInfo.class);
if ((callResponse != null) && (callResponse.getResult() instanceof V1Pod)) {
V1Pod pod = (V1Pod) callResponse.getResult();
Optional.ofNullable(packet.getSpi(DomainPresenceInfo.class))
Optional.ofNullable(info)
.ifPresent(i -> i.setServerPodFromEvent(getPodLabel(pod, LabelConstants.SERVERNAME_LABEL), pod));
}
if (isReady(callResponse.getResult())) {
if (isReady(callResponse.getResult()) || callback.didResume.get()) {
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't understand why we need to check callback.didResume here. How would the operator get here again if the callback has been resumed already?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The check callback.didResume is to avoid the race condition. We do the periodic listing of the introspector job to check if it's ready and also dispatch the callback when the watch event is received. In this scenario, the watch event notifications are flowing. After the intro job completed watch notification is received, the callback is removed and the fiber is resumed. After this, the intro job gets deleted but the child fiber that's listing the job periodically to check for readiness doesn't see the job as ready since it got deleted. Hence the child fiber never finishes and times out. The above check will terminate the child fiber after intro job has been deleted.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If I understand it correctly, check for didResume (didResumeFiber now) is only needed for the job case. If so, we need to remove the check in the pod code path,

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The didResumeFiber will return true only when the fiber is resumed and the resource should be ready at that time. I think there's no harm in keeping the check as that should avoid any potential race condition in pod case as well.

callback.proceedFromWait(callResponse.getResult());
Optional.ofNullable(info).ifPresent(i -> i.resetWatchBackstopRecheckCount());
return doNext(packet);
}
return doDelay(createReadAndIfReadyCheckStep(callback), packet,
getWatchBackstopRecheckDelaySeconds(), TimeUnit.SECONDS);
if (shouldWait(callResponse, info)) {
// Watch backstop recheck count is less than or equal to the configured recheck count, delay.
return doDelay(createReadAndIfReadyCheckStep(callback), packet,
getWatchBackstopRecheckDelaySeconds(), TimeUnit.SECONDS);
} else {
// Watch backstop recheck count is more than configured recheck count, proceed to make-right step.
return doNext(new CallBuilder().readDomainAsync(info.getDomainUid(),
info.getNamespace(), new MakeRightDomainStep(callback, null)), packet);
}
}

private boolean shouldWait(CallResponse<T> callResponse, DomainPresenceInfo info) {
return ((callResponse != null) && (callResponse.getResult() instanceof V1Job))
|| (info == null)
|| (info.incrementAndGetWatchBackstopRecheckCount() <= getWatchBackstopRecheckCount());
}

private String getPodLabel(V1Pod pod, String labelName) {
Expand Down Expand Up @@ -239,6 +266,25 @@ public NextAction apply(Packet packet) {

}

private class MakeRightDomainStep extends DefaultResponseStep {
private final Callback callback;

MakeRightDomainStep(Callback callback, Step next) {
super(next);
this.callback = callback;
}

@Override
public NextAction onSuccess(Packet packet, CallResponse callResponse) {
MakeRightDomainOperation makeRightDomainOperation =
(MakeRightDomainOperation)packet.get(MAKE_RIGHT_DOMAIN_OPERATION);
makeRightDomainOperation.setLiveInfo(new DomainPresenceInfo((Domain)callResponse.getResult()));
callback.fiber.terminate(null, packet);
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

we probably need to call removeCallback here.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I made the change to remove callback here.

makeRightDomainOperation.withExplicitRecheck().interrupt().execute();
return super.onSuccess(packet, callResponse);
}
}

private class Callback implements Consumer<T> {
private final AsyncFiber fiber;
private final Packet packet;
Expand Down
13 changes: 13 additions & 0 deletions operator/src/main/java/oracle/kubernetes/operator/helpers/DomainPresenceInfo.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -55,6 +55,7 @@ public class DomainPresenceInfo {
private final AtomicBoolean isDeleting = new AtomicBoolean(false);
private final AtomicBoolean isPopulated = new AtomicBoolean(false);
private final AtomicInteger retryCount = new AtomicInteger(0);
private final AtomicInteger watchBackstopRecheckCount = new AtomicInteger(0);
private final AtomicReference<Collection<ServerStartupInfo>> serverStartupInfo;
private final AtomicReference<Collection<ServerShutdownInfo>> serverShutdownInfo;

Expand Down Expand Up @@ -559,6 +560,18 @@ int getRetryCount() {
return retryCount.get();
}

public void resetWatchBackstopRecheckCount() {
watchBackstopRecheckCount.set(0);
}

public int incrementAndGetWatchBackstopRecheckCount() {
return watchBackstopRecheckCount.incrementAndGet();
}

public int getWatchBackstopRecheckCount() {
return watchBackstopRecheckCount.get();
}

/** Sets the last completion time to now. */
public void complete() {
resetFailureCount();
Expand Down
2 changes: 1 addition & 1 deletion operator/src/test/java/oracle/kubernetes/operator/WatcherTestBase.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,7 +40,7 @@ public abstract class WatcherTestBase extends ThreadFactoryTestBase implements A
private final List<Memento> mementos = new ArrayList<>();
private final List<Watch.Response<?>> callBacks = new ArrayList<>();
private final AtomicBoolean stopping = new AtomicBoolean(false);
final WatchTuning tuning = new WatchTuning(30, 0, 5);
final WatchTuning tuning = new WatchTuning(30, 0, 5, 24);
private BigInteger resourceVersion = INITIAL_RESOURCE_VERSION;

private V1ObjectMeta createMetaData() {
Expand Down
2 changes: 1 addition & 1 deletion operator/src/test/java/oracle/kubernetes/operator/steps/ManagedServerUpIteratorStepTest.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -95,7 +95,7 @@ public class ManagedServerUpIteratorStepTest extends ThreadFactoryTestBase imple
private final AtomicBoolean stopping = new AtomicBoolean(false);
private static final BigInteger INITIAL_RESOURCE_VERSION = new BigInteger("234");
private final PodWatcher watcher = createWatcher(NS, stopping, INITIAL_RESOURCE_VERSION);
final TuningParameters.WatchTuning tuning = new TuningParameters.WatchTuning(30, 0, 5);
final TuningParameters.WatchTuning tuning = new TuningParameters.WatchTuning(30, 0, 5, 24);

@Nonnull
private static String getManagedServerName(int n) {
Expand Down