-
-
Notifications
You must be signed in to change notification settings - Fork 459
perf(connectivity): Cache network capabilities and status to reduce IPC calls #4560
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.
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
Conversation
} | ||
lastCacheUpdateTime = timeProvider.getCurrentTimeMillis(); | ||
|
||
options |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
could drop this log call, but I thought it might be very useful to investigate potential issues. Under the hood getConnectionStatus and getConnectionType are very lightweight for a cache NetworkCapabilities instance, they just do bitmask operations to get the transport and capability
Performance metrics 🚀
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nice! I've left a nit about the transports, but looks good otherwise!
...core/src/main/java/io/sentry/android/core/internal/util/AndroidConnectionStatusProvider.java
Outdated
Show resolved
Hide resolved
...core/src/main/java/io/sentry/android/core/internal/util/AndroidConnectionStatusProvider.java
Show resolved
Hide resolved
capabilities[0] = NetworkCapabilities.NET_CAPABILITY_INTERNET; | ||
if (buildInfoProvider.getSdkInfoVersion() >= Build.VERSION_CODES.M) { | ||
capabilities[1] = NetworkCapabilities.NET_CAPABILITY_VALIDATED; | ||
} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Bug: Static Array Mutation Causes Network Capability Inconsistency
The static final
capabilities
array is mutated in the AndroidConnectionStatusProvider
constructor based on instance-specific SDK versions. This creates a race condition where multiple instances, particularly those with different API levels, overwrite the shared static array, leading to inconsistent and incorrect network capability checks across all instances.
Locations (1)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
While there might be multiple SDK instances running at a time, it's not possible for them to run on different API levels at the same time, because the device can only target one API level at a time.
* Do not report cached events as lost * E2E tests for OpenTelemetry based console sample (#4563) * e2e tests for console app * fix test failures by waiting for 10s after first try to find envelopes * add system-test-runner.py script to replace bash scripts for running e2e / system tests * use py script for ci, cleanup, makefile * Format code * remove bash scripts * install requests module * api * fix gh script * Implement E2E tests for OTel based console sample * fixes after merge * Format code * e2e tests for console app * Implement E2E tests for OTel based console sample * fixes after merge * Format code * api * Reduce scope forking when using OpenTelemetry (#4565) * Reduce scope forking in OpenTelemetry * Format code * api * changelog --------- Co-authored-by: Sentry Github Bot <bot+github-bot@sentry.io> * SDKs send queue is no longer shutdown immediately on re-init (#4564) * Let queue drain on a restart * Format code * Format code * Update sentry-samples/sentry-samples-console-opentelemetry-noagent/src/test/kotlin/sentry/systemtest/ConsoleApplicationSystemTest.kt * Let queue drain on a restart * Format code * Format code * Update sentry-samples/sentry-samples-console-opentelemetry-noagent/src/test/kotlin/sentry/systemtest/ConsoleApplicationSystemTest.kt * adapt tests * changelog --------- Co-authored-by: Sentry Github Bot <bot+github-bot@sentry.io> --------- Co-authored-by: Sentry Github Bot <bot+github-bot@sentry.io> * release: 8.18.0 * ref(replay): Use main thread to schedule capture (#4542) * perf(connectivity): Cache network capabilities and status to reduce IPC calls (#4560) * fix(breadcrumbs): Deduplicate battery breadcrumbs (#4561) * fix(ci): remove obsolete NDK debug symbols (#4581) As they don't exist anymore and this is done within sentry-native directly: https://github.com/getsentry/sentry-native/pull/1327/files * fix(android): Remove unused method (#4585) * fix(android): Remove unused method * Update Changelog * Add rules file for documenting SDK offline behaviour (#4572) #skip-changelog ## 📜 Description <!--- Describe your changes in detail --> Add rules file for documenting SDK offline behaviour ## 💡 Motivation and Context <!--- Why is this change required? What problem does it solve? --> <!--- If it fixes an open issue, please link to the issue here. --> Should help speed up AI reasoning about the SDK offline/retry behaviour. ## 💚 How did you test it? ## 📝 Checklist <!--- Put an `x` in the boxes that apply --> - [ ] I added tests to verify the changes. - [ ] No new PII added or SDK only sends newly added PII if `sendDefaultPII` is enabled. - [ ] I updated the docs if needed. - [ ] I updated the wizard if needed. - [ ] Review from the native team if needed. - [ ] No breaking change or entry added to the changelog. - [ ] No breaking change for hybrid SDKs or communicated to hybrid SDKs. ## 🔮 Next steps * perf(connectivity): Have only one NetworkCallback active at a time (#4562) * fix(scripts): update-gradle script set-version (#4591) * fix: sentry-android-ndk proguard rule keeps all native class (#4427) * fix: sentry-androi-ndk proguard rule keeps all native class * docs: update CHANGELOG * fix: update CHANGELOG * Update CHANGELOG.md * Update CHANGELOG.md --------- Co-authored-by: Markus Hintersteiner <m.hintersteiner@gmail.com> Co-authored-by: Markus Hintersteiner <markus.hintersteiner@sentry.io> * refactor(lifecycle): Use single lifecycle observer (#4567) * perf(connectivity): Cache network capabilities and status to reduce IPC calls * changelog * Changelog * revert * fix(breadcrumbs): Deduplicate battery breadcrumbs * ref * Changelog * Fix test * perf(connectivity): Have only one NetworkCallback active at a time * Changelog * perf(integrations): Use single lifecycle observer * Add tests * Changelog * Fix tests * Improve callback handling and test visibility (#4593) * Null-check lifecycleObserver --------- Co-authored-by: Markus Hintersteiner <markus.hintersteiner@sentry.io> * fix(sqlite): Fix abstract method error (#4597) * fix(sqlite): Fix abstract method error * Update CHANGELOG.md * Suppress metadata version checks * perf(integrations): Do not register for SystemEvents and NetworkCallbacks when launched with background importance (#4579) * fix(android): Ensure frame metrics listeners are registered/unregistered on the main thread (#4582) * fix(android): Ensure frame metrics listeners are registered/unregistered on the main thread * Fix race conditions * Update Changelog * Update CHANGELOG.md * Address PR feedback * perf(executor): Prewarm SentryExecutorService (#4606) * review feedback * changelog * pass through whether cache stored in AndroidEnvelopeCache + test * Format code --------- Co-authored-by: Sentry Github Bot <bot+github-bot@sentry.io> Co-authored-by: getsentry-bot <bot@sentry.io> Co-authored-by: Roman Zavarnitsyn <rom4ek93@gmail.com> Co-authored-by: Markus Hintersteiner <markus.hintersteiner@sentry.io> Co-authored-by: Ghasem Shirdel <shirdelghasem79@gmail.com> Co-authored-by: Markus Hintersteiner <m.hintersteiner@gmail.com>
📜 Description
onLost
/onUnavailable
would immediately send DISCONNECTED now as opposed to retrievingconnectionStatus
againisSignificantChange
check, becauseonNetworkCapabilitiesChanged
can be called many times due to bandwidth change/etc. and we're only interested in transport/connectivity changesRan a test with perfetto to count the number of binder calls, and these are the results:
Before (83 binder calls from SentryExecService)
After (5 binder calls from SentryExecService)
💡 Motivation and Context
Fixes #4200
💚 How did you test it?
Manually + automated
📝 Checklist
sendDefaultPII
is enabled.🔮 Next steps