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

Turn off Statsbeat when proxy is used or any exception from the server #2221

Merged
merged 17 commits into from
Apr 13, 2022
Merged

Turn off Statsbeat when proxy is used or any exception from the server #2221

Changes from 1 commit
Commits
Show all changes
17 commits
Select commit Hold shift + click to select a range
3fbf7bc
Turn off statsbeat when proxy used or any exception from the server
heyams Apr 7, 2022
5ed97bf
Fix comment format
heyams Apr 7, 2022
59c17dc
Add a todo
heyams Apr 7, 2022
e53cd2f
Address feedback
heyams Apr 7, 2022
b21e433
Change todo
heyams Apr 8, 2022
1652a6e
Shut down after 3 statsbeat errors
heyams Apr 8, 2022
9b097d9
Reset statsbeat error count on success
heyams Apr 8, 2022
de1e052
Use one time success flag and increment max to 10
heyams Apr 8, 2022
0a2e23c
No need to reset
heyams Apr 9, 2022
a8dae0f
Suppress statbeat error logging
trask Apr 9, 2022
539dca9
Simplify and avoid other unknown cases
trask Apr 9, 2022
9f9050d
move and comment
trask Apr 9, 2022
22ef957
one more logging
trask Apr 9, 2022
1598512
Update to latest spec PR
trask Apr 12, 2022
5f6a1c0
imports
trask Apr 13, 2022
a131213
Add comment
trask Apr 13, 2022
0dfa662
Add TODO
trask Apr 13, 2022
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
36 changes: 20 additions & 16 deletions ...ain/java/com/microsoft/applicationinsights/agent/internal/telemetry/TelemetryChannel.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,6 +52,7 @@
import java.nio.ByteBuffer;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.HashSet;
import java.util.List;
import java.util.Map;
import java.util.concurrent.atomic.AtomicBoolean;
Expand All @@ -73,19 +74,17 @@ public class TelemetryChannel {

private static final AppInsightsByteBufferPool byteBufferPool = new AppInsightsByteBufferPool();

private static final int MAX_STATSBEAT_ERROR_COUNT = 10;

private final OperationLogger operationLogger;
private final OperationLogger retryOperationLogger;

// TODO (kryalama) do we still need this AtomicBoolean, or can we use throttling built in to the
// operationLogger?
private final AtomicBoolean friendlyExceptionThrown = new AtomicBoolean();

private final AtomicInteger statsbeatErrorCount = new AtomicInteger();
private final AtomicBoolean statsbeatShutdown = new AtomicBoolean();
private final AtomicInteger statsbeatUnableToReachBreezeCounter = new AtomicInteger();
trask marked this conversation as resolved.
Show resolved Hide resolved
private final AtomicBoolean statsbeatHasBeenShutdown = new AtomicBoolean();
trask marked this conversation as resolved.
Show resolved Hide resolved

private volatile boolean statsbeatAtLeastOneSuccess;
private volatile boolean statsbeatHasReachedBreezeAtLeastOnce;
trask marked this conversation as resolved.
Show resolved Hide resolved

@SuppressWarnings("CatchAndPrintStackTrace")
private static ObjectMapper createObjectMapper() {
Expand Down Expand Up @@ -122,7 +121,7 @@ public CompletableResultCode sendRawBytes(
String instrumentationKey,
Runnable onSuccess,
Consumer<Boolean> onFailure) {
if (isStatsbeat && statsbeatShutdown.get()) {
if (isStatsbeat && statsbeatHasBeenShutdown.get()) {
trask marked this conversation as resolved.
Show resolved Hide resolved
// let it be deleted from disk so that it won't keep getting retried
return CompletableResultCode.ofSuccess();
}
Expand Down Expand Up @@ -299,6 +298,11 @@ private CompletableResultCode internalSend(
return result;
}

// not including 401/403/503 in this list because those are commonly returned by proxy servers
// when they are not configured to allow traffic for westus-0
private static final Set<Integer> RESPONSE_CODES_INDICATING_REACHED_BREEZE =
new HashSet<>(asList(200, 206, 402, 408, 429, 439, 500));
trask marked this conversation as resolved.
Show resolved Hide resolved
trask marked this conversation as resolved.
Show resolved Hide resolved

private Consumer<HttpResponse> responseHandler(
String instrumentationKey,
long startTime,
Expand All @@ -313,11 +317,11 @@ private Consumer<HttpResponse> responseHandler(
.subscribe(
body -> {
int statusCode = response.getStatusCode();
if (isStatsbeat && !statsbeatAtLeastOneSuccess) {
if (statusCode == 200) {
statsbeatAtLeastOneSuccess = true;
} else if (statusCode != 439 && statusCode != 402) {
possiblyShutDownStatsbeat();
if (isStatsbeat && !statsbeatHasReachedBreezeAtLeastOnce) {
if (RESPONSE_CODES_INDICATING_REACHED_BREEZE.contains(statusCode)) {
statsbeatHasReachedBreezeAtLeastOnce = true;
} else {
statsbeatDidNotReachBreeze();
}
}
switch (statusCode) {
Expand Down Expand Up @@ -388,8 +392,8 @@ private Consumer<Throwable> errorHandler(
String instrumentationKey, Consumer<Boolean> onFailure, OperationLogger operationLogger) {

return error -> {
if (isStatsbeat && !statsbeatAtLeastOneSuccess) {
possiblyShutDownStatsbeat();
if (isStatsbeat && !statsbeatHasReachedBreezeAtLeastOnce) {
statsbeatDidNotReachBreeze();
}

if (!isStatsbeat
Expand All @@ -407,9 +411,9 @@ private Consumer<Throwable> errorHandler(
};
}

private void possiblyShutDownStatsbeat() {
if (statsbeatErrorCount.getAndIncrement() >= MAX_STATSBEAT_ERROR_COUNT
&& !statsbeatShutdown.getAndSet(true)) {
private void statsbeatDidNotReachBreeze() {
if (statsbeatUnableToReachBreezeCounter.getAndIncrement() >= 10
&& !statsbeatHasBeenShutdown.getAndSet(true)) {
trask marked this conversation as resolved.
Show resolved Hide resolved
// shutting down statsbeat because it's unlikely that it will ever get through at this point
// some possible reasons:
// * AMPLS
Expand Down