according to Naveen tests pass locally, issue in downloading the emulator.
* mirror of https://github.com/Azure/azure-cosmosdb-java/pull/395
* need to rerun the ci again after agent upgrade from ububtu 16 to 20
* need to rerun the ci again after agent upgrade from ububtu 16 to 20
* adding space to trigger ci which is in struct state
Co-authored-by: Mohammad Derakhshani <moderakh@users.noreply.github.com>
* Removed generic IO exception condition in network failure. Replaced with specific exceptions. Included in unit testing
* Updated code owners for this repo
* Fixing reagion failover issue for direct tcp calls
* indentation
* adding test case and covering issue scenerio for writes
* Fixing lazy indexing tests and wrapping ex to cosmos exception for network test
* resolving comments
* indentation change
* resolving comment
* Tidied a couple of code files and add some code comments
* Addressed memory leak by simplifying reference counting mechanism. The e2e test passes without leaks being detected by the paranoid memory leak checker. More testing to follow.
* Removed some commented-out code and tweaked a checkNotNull error message
* Addressed a compilation issue
Co-authored-by: Kushagra Thapar <kuthapar@microsoft.com>
* DocumentClientException is now thread-safe (but offers no consistency guarantees) and we have a test that request/response headers can be enumerated while they're being updated (sans ConcurrentModificationExceptions).
* Bug fix: Adjust for fact that ConcurrentMap does not accept the null values that we sometimes pass.
* Added two new DocumentClientExceptionTest methods and increased the timeout interval on ConsistencyTest2.validateSessionTokenAsync because it times out regularly in local and CI test runs with the old timeout interval.
* Configs.getProtocol now accepts in priority order -Dcosmos.directModeProtocol, -Dazure.cosmos.directModeProtocol (for compatability with v4) and the environment variable DIRECT_MODE_PROTOCOL (which is sometimes convenient in test environments)
* ReadMyWritesConsistencyTest now sets document data field size and RntbdTransportClient can now load default direct tcp options
* Removed option to specify document data field size on ReadMyWritesConsistencyTest and fixed a test break.
* Enclosed static initializer in try/finally to reduce chance of unexpected failures and renamed a mvn version property for clarity.
* Bug fix and tweaks based on code review feedback
* Increased request timer resolution from 5 to 100 ms
* Decreased request timer resolution from 100 to 5 ms
* Added RntbdTransportClient.Options.requestTimerResolution property for use by RntbdRequestTimer and renamed long times with unit suffix (e.g., InNanos for nanosecond values and InMillis for millisecond values)
* Minor code cleanup for compliance with central sdk standards
* Updated javadocs
* Updated sdk/cosmos/README.md with info on using system properties to modify default Direct TCP options
* Added RntbdMetrics.requestSize and RntbdMetrics.responseSize which track the distribution of Direct TCP request and response messages. Also: ported RequestTimeline class from v4 and tidied the code in touched files to eliminate checkstyle errors.
Co-authored-by: Kushagra Thapar <kushuthapar@gmail.com>
* change thread to avoid deserialization to happen on io-thread
* build and install to local m2 script
* update
* change io thread in GatewayStoreModel
* Configs.getProtocol now accepts in priority order -Dcosmos.directModeProtocol, -Dazure.cosmos.directModeProtocol (for compatability with v4) and the environment variable DIRECT_MODE_PROTOCOL (which is sometimes convenient in test environments)
* ReadMyWritesConsistencyTest now sets document data field size and RntbdTransportClient can now load default direct tcp options
* Removed option to specify document data field size on ReadMyWritesConsistencyTest and fixed a test break.
* Enclosed static initializer in try/finally to reduce chance of unexpected failures and renamed a mvn version property for clarity.
* Bug fix and tweaks based on code review feedback
* Increased request timer resolution from 5 to 100 ms
* Decreased request timer resolution from 100 to 5 ms
* Added RntbdTransportClient.Options.requestTimerResolution property for use by RntbdRequestTimer and renamed long times with unit suffix (e.g., InNanos for nanosecond values and InMillis for millisecond values)
* Minor code cleanup for compliance with central sdk standards
* Updated javadocs
* Updated sdk/cosmos/README.md with info on using system properties to modify default Direct TCP options
Co-authored-by: Kushagra Thapar <kushuthapar@gmail.com>
* Configs.getProtocol now accepts in priority order -Dcosmos.directModeProtocol, -Dazure.cosmos.directModeProtocol (for compatability with v4) and the environment variable DIRECT_MODE_PROTOCOL (which is sometimes convenient in test environments)
* ReadMyWritesConsistencyTest now sets document data field size and RntbdTransportClient can now load default direct tcp options
* Removed option to specify document data field size on ReadMyWritesConsistencyTest and fixed a test break.
* Enclosed static initializer in try/finally to reduce chance of unexpected failures and renamed a mvn version property for clarity.
* Bug fix and tweaks based on code review feedback
* Increased request timer resolution from 5 to 100 ms
* Decreased request timer resolution from 100 to 5 ms
* Added RntbdTransportClient.Options.requestTimerResolution property for use by RntbdRequestTimer and renamed long times with unit suffix (e.g., InNanos for nanosecond values and InMillis for millisecond values)
* Corrected an error in RequestRateTooLargeException, a regression that caused the status code to be set incorrectly
* Corrected ServiceUnavailableException status code, optimized imports, and added DocumentClientExceptionTest.statusCodeIsCorrect
The latter resulted in moving DocuemntClientExceptionTest from commons to direct-impl. Without the move commons would need to take a dependency on direct-impl which seems undesirable.
* Refactored DocumentClientExceptionTest to spread it out between commons and direct-impl
* Bumped version number to 2.6.3-SNAPSHOT
* Bumped netty version number to pickup this and other fixes: https://github.com/netty/netty/pull/9575
* Improved error handling for RntbdResponseHeaders
* Improved an error message
* Resolved#275
* Minor code cleanup in the vicinity of one possible source of this issue
* Tweaks
* Added logger.info that reports the ssl provider (OPENSSL or JVM)
* Improved diagnostics
* Added state tracking to RntbdRequestRecord so that the logs reveal the state of a request when it times out
* Tweaked text of RequestTimeoutException produced by RntbdRequestRecord.expire
* Added RntbdRequestRecord.State.UNSENT to denote failed writes
* Sorted methods and improved checkArgument messages
* Tweaks
* Added logger for tracking hashed wheel timer stops
* Added debug logger message
* Tweaked comments and line spacing
* Added final modifier
* Added requestExpiryInterval to RntbdTransportClient.Options
* Version bump
* Tweaks for performance testing
* Port from azure-cosmos-4.X
* Removed extraneous file
* Updated changelog to force a rebuild
* Fixed a test break
* Corrected package names in logger configuration file
* Updated changelog that tracks unreleased changes
* Corrected an error in RequestRateTooLargeException, a regression that caused the status code to be set incorrectly
* Corrected ServiceUnavailableException status code, optimized imports, and added DocumentClientExceptionTest.statusCodeIsCorrect
The latter resulted in moving DocuemntClientExceptionTest from commons to direct-impl. Without the move commons would need to take a dependency on direct-impl which seems undesirable.
* Refactored DocumentClientExceptionTest to spread it out between commons and direct-impl
* Bumped version number to 2.6.3-SNAPSHOT
* Bumped netty version number to pickup this and other fixes: https://github.com/netty/netty/pull/9575
* Improved error handling for RntbdResponseHeaders
* Improved an error message
* Corrected an error in RequestRateTooLargeException, a regression that caused the status code to be set incorrectly
* Corrected ServiceUnavailableException status code, optimized imports, and added DocumentClientExceptionTest.statusCodeIsCorrect
The latter resulted in moving DocuemntClientExceptionTest from commons to direct-impl. Without the move commons would need to take a dependency on direct-impl which seems undesirable.
* Refactored DocumentClientExceptionTest to spread it out between commons and direct-impl
* Bumped version number to 2.6.3-SNAPSHOT
* version bump to 2.4.6-SNAPSHOT
* byte buffer pool optimized
* unnessery uuid conversion
* URI changed to Uri
* fixes
* initlize StringBuilder with proper size for auth token
* minor fix in eralier commits
* fixed test
* addressed review comments
* fix in reading bytes set limit, fixed test
* fixed a broken test
* some unrelated test failrue: fixed
* fixed tests
* 2.4.6 release
* added getters for Uri and made fields final
Aayush Kataria
Naveen Singh
Kushagra Thapar
Bhaskar Mallapragada
David Noble
Bhaskar Mallapragada
Mohammad Derakhshani
Christopher Anderson