AWS for Fluent Bit 2.31.5

2.31.5

This release includes:

• Fluent Bit 1.9.10
• Amazon CloudWatch Logs for Fluent Bit 1.9.1
• Amazon Kinesis Streams for Fluent Bit 1.10.1
• Amazon Kinesis Firehose for Fluent Bit 1.7.1

Compared to 2.23.4 this release adds:

• Feature - In the init image, extract Availability Zone from ECS Task Metadata and inject into Fluent Bit container as AWS_AVAILABILITY_ZONE environment variable aws-for-fluent-bit:539

Compared to 2.31.4, this release removes the following fixes that may introduce instabilities to the S3 plugin:

• Enhancement - Transition S3 to fully synchronous file uploads to improve plugin stability fluent-bit:6573

Same as 2.31.4, this release removes the following fixes that may introduce instabilities to the S3 plugin:

• Bug - Support Retry_Limit option in S3 plugin to set retries fluent-bit:6475
• Bug - Format S3 filename with timestamp from the first log in uploaded file, rather than the time the first log was buffered by the s3 output aws-for-fluent-bit:459

Same as 2.31.4, this release includes the following fixes and features that we are working on getting accepted upstream:

• Bug - Resolve cloudwatch_logs duplicate tag match SIGSEGV issue introduced in 2.29.0 aws-for-fluent-bit:542
• Feature - Add kinesis_firehose and kinesis_streams support for time_key_format milliseconds with %3N option, and nanoseconds 9N and %L options fluent-bit:2831
• Feature - Support OpenSearch Serverless data ingestion via OpenSearch plugin fluent-bit:6448
• Bug - Mitigate Datadog output plugin issue by reverting recent PR aws-for-fluent-bit:491
• Bug - Resolve S3 logic to display log_key missing warning message if the configured log_key field is missing from log payload fluent-bit:6557
• Bug - ECS Metadata filter gracefuly handle task metadata query errors and cache metadata processing state to improve performance aws-for-fluent-bit:505
• Bug - Resolve a rare Datadog segfault that occurs when remapping tags aws-for-fluent-bit:491
• Bug - Resolve net.keepalive SIGSEGV issue and priority event loop corruption bug aws-for-fluent-bit:542

---

We’ve run the new released image in our ECS load testing framework and here is the result. This testing result provides benchmarks of aws-for-fluent-bit under different input load. Learn more about the load test.

plugin	source		20 MB/s	25 MB/s	30 MB/s
kinesis_firehose	stdstream	Log Loss	✅	✅	✅
		Log Duplication	0%(31089)	0%(6000)	✅
kinesis_firehose	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
kinesis_streams	stdstream	Log Loss	✅	✅	0%(43769)
		Log Duplication	0%(13562)	0%(45658)	0%(108603)
kinesis_streams	tcp	Log Loss	✅	✅	✅
		Log Duplication	0%(1500)	✅	0%(27809)
s3	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
s3	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅

plugin	source		1 MB/s	2 MB/s	3 MB/s
cloudwatch_logs	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
cloudwatch_logs	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅

AWS for Fluent Bit 2.31.4

2.31.4

This release includes:

• Fluent Bit 1.9.10
• Amazon CloudWatch Logs for Fluent Bit 1.9.1
• Amazon Kinesis Streams for Fluent Bit 1.10.1
• Amazon Kinesis Firehose for Fluent Bit 1.7.1

Compared to 2.31.3, this release removes the following fixes that may introduce instabilities to the S3 plugin:

• Bug - Support Retry_Limit option in S3 plugin to set retries fluent-bit:6475
• Bug - Format S3 filename with timestamp from the first log in uploaded file, rather than the time the first log was buffered by the s3 output aws-for-fluent-bit:459

Same as 2.31.3, this release includes the following fixes and features that we are working on getting accepted upstream:

• Bug - Resolve cloudwatch_logs duplicate tag match SIGSEGV issue introduced in 2.29.0 aws-for-fluent-bit:542
• Feature - Add kinesis_firehose and kinesis_streams support for time_key_format milliseconds with %3N option, and nanoseconds 9N and %L options fluent-bit:2831
• Feature - Support OpenSearch Serverless data ingestion via OpenSearch plugin fluent-bit:6448
• Enhancement - Transition S3 to fully synchronous file uploads to improve plugin stability fluent-bit:6573
• Bug - Mitigate Datadog output plugin issue by reverting recent PR aws-for-fluent-bit:491
• Bug - Resolve S3 logic to display log_key missing warning message if the configured log_key field is missing from log payload fluent-bit:6557
• Bug - ECS Metadata filter gracefuly handle task metadata query errors and cache metadata processing state to improve performance aws-for-fluent-bit:505
• Bug - Resolve a rare Datadog segfault that occurs when remapping tags aws-for-fluent-bit:491
• Bug - Resolve net.keepalive SIGSEGV issue and priority event loop corruption bug aws-for-fluent-bit:542

---

We’ve run the new released image in our ECS load testing framework and here is the result. This testing result provides benchmarks of aws-for-fluent-bit under different input load. Learn more about the load test.

plugin	source		20 MB/s	25 MB/s	30 MB/s
kinesis_firehose	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
kinesis_firehose	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
kinesis_streams	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	0%(18360)	0%(32342)
kinesis_streams	tcp	Log Loss	✅	✅	0%(24)
		Log Duplication	0%(500)	✅	0%(26068)
s3	stdstream	Log Loss	✅	7%(1085784)	27%(4931846)
		Log Duplication	✅	✅	✅
s3	tcp	Log Loss	✅	✅	14%(2636934)
		Log Duplication	✅	✅	✅

plugin	source		1 MB/s	2 MB/s	3 MB/s
cloudwatch_logs	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
cloudwatch_logs	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅

AWS for Fluent Bit 2.31.3

2.31.3

This release includes:

• Fluent Bit 1.9.10
• Amazon CloudWatch Logs for Fluent Bit 1.9.1
• Amazon Kinesis Streams for Fluent Bit 1.10.1
• Amazon Kinesis Firehose for Fluent Bit 1.7.1

Compared to 2.31.2, this release adds the following fix that we are working on getting accepted upstream:

• Bug - Resolve net.keepalive SIGSEGV issue and priority event loop corruption bug aws-for-fluent-bit:542

Same as 2.31.2, this release includes the following fixes and features that we are working on getting accepted upstream:

• Bug - Resolve cloudwatch_logs duplicate tag match SIGSEGV issue introduced in 2.29.0 aws-for-fluent-bit:542
• Feature - Add kinesis_firehose and kinesis_streams support for time_key_format milliseconds with %3N option, and nanoseconds 9N and %L options fluent-bit:2831
• Feature - Support OpenSearch Serverless data ingestion via OpenSearch plugin fluent-bit:6448
• Enhancement - Transition S3 to fully synchronous file uploads to improve plugin stability fluent-bit:6573
• Bug - Mitigate Datadog output plugin issue by reverting recent PR aws-for-fluent-bit:491
• Bug - Format S3 filename with timestamp from the first log in uploaded file, rather than the time the first log was buffered by the s3 output aws-for-fluent-bit:459
• Bug - Resolve S3 logic to display log_key missing warning message if the configured log_key field is missing from log payload fluent-bit:6557
• Bug - ECS Metadata filter gracefuly handle task metadata query errors and cache metadata processing state to improve performance aws-for-fluent-bit:505
• Bug - Support Retry_Limit option in S3 plugin to set retries fluent-bit:6475
• Bug - Resolve a rare Datadog segfault that occurs when remapping tags aws-for-fluent-bit:491

---

We’ve run the new released image in our ECS load testing framework and here is the result. This testing result provides benchmarks of aws-for-fluent-bit under different input load. Learn more about the load test.

plugin	source		20 MB/s	25 MB/s	30 MB/s
kinesis_firehose	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	0%(932)
kinesis_firehose	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
kinesis_streams	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	0%(9098)	0%(26069)
kinesis_streams	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	0%(1500)	0%(25373)
s3	stdstream	Log Loss	✅	12%(1862790)	23%(4223974)
		Log Duplication	✅	✅	✅
s3	tcp	Log Loss	✅	✅	15%(2783857)
		Log Duplication	✅	✅	✅

plugin	source		1 MB/s	2 MB/s	3 MB/s
cloudwatch_logs	stdstream	Log Loss	✅	✅	11%(207189)
		Log Duplication	✅	✅	✅
cloudwatch_logs	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅

AWS for Fluent Bit 2.31.2

2.31.2

This release includes:

• Fluent Bit 1.9.10
• Amazon CloudWatch Logs for Fluent Bit 1.9.1
• Amazon Kinesis Streams for Fluent Bit 1.10.1
• Amazon Kinesis Firehose for Fluent Bit 1.7.1

Compared to 2.31.1, this release adds the following feature that we are working on getting accepted upstream:

• Bug - Resolve cloudwatch_logs duplicate tag match SIGSEGV issue introduced in 2.29.0 aws-for-fluent-bit:542

Same as 2.31.1, this release includes the following fixes and features that we are working on getting accepted upstream:

• Feature - Add kinesis_firehose and kinesis_streams support for time_key_format milliseconds with %3N option, and nanoseconds 9N and %L options fluent-bit:2831
• Feature - Support OpenSearch Serverless data ingestion via OpenSearch plugin fluent-bit:6448
• Enhancement - Transition S3 to fully synchronous file uploads to improve plugin stability fluent-bit:6573
• Bug - Mitigate Datadog output plugin issue by reverting recent PR aws-for-fluent-bit:491
• Bug - Format S3 filename with timestamp from the first log in uploaded file, rather than the time the first log was buffered by the s3 output aws-for-fluent-bit:459
• Bug - Resolve S3 logic to display log_key missing warning message if the configured log_key field is missing from log payload fluent-bit:6557
• Bug - ECS Metadata filter gracefuly handle task metadata query errors and cache metadata processing state to improve performance aws-for-fluent-bit:505
• Bug - Support Retry_Limit option in S3 plugin to set retries fluent-bit:6475
• Bug - Resolve a rare Datadog segfault that occurs when remapping tags aws-for-fluent-bit:491

---

We’ve run the new released image in our ECS load testing framework and here is the result. This testing result provides benchmarks of aws-for-fluent-bit under different input load. Learn more about the load test.

plugin	source		20 MB/s	25 MB/s	30 MB/s
kinesis_firehose	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
kinesis_firehose	tcp	Log Loss	✅	✅	✅
		Log Duplication	0%(500)	✅	✅
kinesis_streams	stdstream	Log Loss	✅	✅	0%(5795)
		Log Duplication	0%(500)	0%(32694)	0%(63482)
kinesis_streams	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	0%(500)	0%(6445)
s3	stdstream	Log Loss	✅	12%(1874773)	30%(5479384)
		Log Duplication	✅	✅	✅
s3	tcp	Log Loss	✅	✅	15%(2818466)
		Log Duplication	✅	✅	✅

plugin	source		1 MB/s	2 MB/s	3 MB/s
cloudwatch_logs	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
cloudwatch_logs	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅

Note:

• The green check ✅ in the table means no log loss or no log duplication.
• Number in parentheses means the number of records out of total records. For example, 0%(1064/1.8M) under 30Mb/s throughput means 1064 duplicate records out of 18M input records by which log duplication percentage is 0%.
• Log loss is the percentage of data lost and log duplication is the percentage of duplicate logs received at the destination. Your results may differ because they can be influenced by many factors like different configs and environment settings. Log duplication is exclusively caused by partially succeeded batches that were retried, which means it is random.

AWS for Fluent Bit 2.31.1

2.31.1

This release includes:

• Fluent Bit 1.9.10
• Amazon CloudWatch Logs for Fluent Bit 1.9.1
• Amazon Kinesis Streams for Fluent Bit 1.10.1
• Amazon Kinesis Firehose for Fluent Bit 1.7.1

Compared to 2.31.0, this release adds the following feature that we are working on getting accepted upstream:

• Bug - Support Retry_Limit option in S3 plugin to set retries fluent-bit:6475
• Bug - Resolve a rare Datadog segfault that occurs when remapping tags aws-for-fluent-bit:491

Same as 2.31.0, this release includes the following fixes and features that we are working on getting accepted upstream:

• Feature - Add kinesis_firehose and kinesis_streams support for time_key_format milliseconds with %3N option, and nanoseconds 9N and %L options fluent-bit:2831
• Feature - Support OpenSearch Serverless data ingestion via OpenSearch plugin fluent-bit:6448
• Enhancement - Transition S3 to fully synchronous file uploads to improve plugin stability fluent-bit:6573
• Bug - Mitigate Datadog output plugin issue by reverting recent PR aws-for-fluent-bit:491
• Bug - Format S3 filename with timestamp from the first log in uploaded file, rather than the time the first log was buffered by the s3 output aws-for-fluent-bit:459
• Bug - Resolve S3 logic to display log_key missing warning message if the configured log_key field is missing from log payload fluent-bit:6557
• Bug - ECS Metadata filter gracefuly handle task metadata query errors and cache metadata processing state to improve performance aws-for-fluent-bit:505

---

We’ve run the new released image in our ECS load testing framework and here is the result. This testing result provides benchmarks of aws-for-fluent-bit under different input load. Learn more about the load test.

plugin	source		20 MB/s	25 MB/s	30 MB/s
kinesis_firehose	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
kinesis_firehose	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
kinesis_streams	stdstream	Log Loss	✅	✅	0%(847)
		Log Duplication	✅	0%(23517)	0%(51518)
kinesis_streams	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
s3	stdstream	Log Loss	0%(20000)	16%(2525305)	29%(5276105)
		Log Duplication	✅	✅	✅
s3	tcp	Log Loss	✅	✅	16%(3000958)
		Log Duplication	✅	✅	✅

plugin	source		1 MB/s	2 MB/s	3 MB/s
cloudwatch_logs	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
cloudwatch_logs	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅

Note:

• The green check ✅ in the table means no log loss or no log duplication.
• Number in parentheses means the number of records out of total records. For example, 0%(1064/1.8M) under 30Mb/s throughput means 1064 duplicate records out of 18M input records by which log duplication percentage is 0%.
• Log loss is the percentage of data lost and log duplication is the percentage of duplicate logs received at the destination. Your results may differ because they can be influenced by many factors like different configs and environment settings. Log duplication is exclusively caused by partially succeeded batches that were retried, which means it is random.

AWS for Fluent Bit 2.31.0

2.31.0

This release includes:

• Fluent Bit 1.9.10
• Amazon CloudWatch Logs for Fluent Bit 1.9.1
• Amazon Kinesis Streams for Fluent Bit 1.10.1
• Amazon Kinesis Firehose for Fluent Bit 1.7.1

Compared to 2.30.0, this release adds the following feature that we are working on getting accepted upstream:

• Feature - Add kinesis_firehose and kinesis_streams support for time_key_format milliseconds with %3N option, and nanoseconds 9N and %L options fluent-bit:2831
• Bug - Format S3 filename with timestamp from the first log in uploaded file, rather than the time the first log was buffered by the s3 output aws-for-fluent-bit:459
• Enhancement - Transition S3 to fully synchronous file uploads to improve plugin stability fluent-bit:6573
• Bug - Resolve S3 logic to display log_key missing warning message if the configured log_key field is missing from log payload fluent-bit:6557
• Bug - ECS Metadata filter gracefuly handle task metadata query errors and cache metadata processing state to improve performance aws-for-fluent-bit:505

Same as 2.30.0, this release includes the following fixes and features that we are working on getting accepted upstream:

• Feature - Support OpenSearch Serverless data ingestion via OpenSearch plugin fluent-bit:6448
• Bug - Mitigate Datadog output plugin issue by reverting recent PR aws-for-fluent-bit:491

---

We’ve run the new released image in our ECS load testing framework and here is the result. This testing result provides benchmarks of aws-for-fluent-bit under different input load. Learn more about the load test.

plugin	source		20 MB/s	25 MB/s	30 MB/s
kinesis_firehose	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
kinesis_firehose	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	0%(946)
kinesis_streams	stdstream	Log Loss	✅	✅	0%(83093)
		Log Duplication	✅	0%(13294)	2%(495013)
kinesis_streams	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	0%(26270)
s3	stdstream	Log Loss	✅	3%(541520)	29%(5305665)
		Log Duplication	✅	✅	✅
s3	tcp	Log Loss	✅	✅	9%(1657116)
		Log Duplication	✅	✅	✅

plugin	source		1 MB/s	2 MB/s	3 MB/s
cloudwatch_logs	stdstream	Log Loss	✅	16%(198844)	✅
		Log Duplication	✅	✅	✅
cloudwatch_logs	tcp	Log Loss	7%(43686)	✅	✅
		Log Duplication	✅	✅	✅

Note:

• The green check ✅ in the table means no log loss or no log duplication.
• Number in parentheses means the number of records out of total records. For example, 0%(1064/1.8M) under 30Mb/s throughput means 1064 duplicate records out of 18M input records by which log duplication percentage is 0%.
• Log loss is the percentage of data lost and log duplication is the percentage of duplicate logs received at the destination. Your results may differ because they can be influenced by many factors like different configs and environment settings. Log duplication is exclusively caused by partially succeeded batches that were retried, which means it is random.

AWS for Fluent Bit 2.30.0

2.30.0

This release includes:

• Fluent Bit 1.9.10
• Amazon CloudWatch Logs for Fluent Bit 1.9.1
• Amazon Kinesis Streams for Fluent Bit 1.10.1
• Amazon Kinesis Firehose for Fluent Bit 1.7.1

Compared to 2.29.1, this release adds the following feature that we are working on getting accepted upstream:

• Feature - Support OpenSearch Serverless data ingestion via OpenSearch plugin fluent-bit:6448

Same as 2.29.1, this release includes the following fix that we are working on getting accepted upstream:

• Bug - Mitigate Datadog output plugin issue by reverting recent PR aws-for-fluent-bit:491

---

We’ve run the new released image in our ECS load testing framework and here is the result. This testing result provides benchmarks of aws-for-fluent-bit under different input load. Learn more about the load test.

plugin	source		20 MB/s	25 MB/s	30 MB/s
kinesis_firehose	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
kinesis_firehose	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	0%(3239)
kinesis_streams	stdstream	Log Loss	✅	✅	0%(3239)
		Log Duplication	✅	0%(18526)	0%(37223)
kinesis_streams	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	0%(2969)	✅
s3	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
s3	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅

plugin	source		1 MB/s	2 MB/s	3 MB/s
cloudwatch_logs	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
cloudwatch_logs	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅

Note:

• The green check ✅ in the table means no log loss or no log duplication.
• Number in parentheses means the number of records out of total records. For example, 0%(1064/1.8M) under 30Mb/s throughput means 1064 duplicate records out of 18M input records by which log duplication percentage is 0%.
• Log loss is the percentage of data lost and log duplication is the percentage of duplicate logs received at the destination. Your results may differ because they can be influenced by many factors like different configs and environment settings. Log duplication is exclusively caused by partially succeeded batches that were retried, which means it is random.

AWS for Fluent Bit 2.29.1

2.29.1

This release includes:

• Fluent Bit 1.9.10
• Amazon CloudWatch Logs for Fluent Bit 1.9.1
• Amazon Kinesis Streams for Fluent Bit 1.10.1
• Amazon Kinesis Firehose for Fluent Bit 1.7.1

This release includes the following fixes for AWS customers that we are working on getting accepted upstream:

• Bug - Mitigate Datadog output plugin issue by reverting recent PR aws-for-fluent-bit:491

---

We’ve run the new released image in our ECS load testing framework and here is the result. This testing result provides benchmarks of aws-for-fluent-bit under different input load. Learn more about the load test.

plugin	source		20 MB/s	25 MB/s	30 MB/s
kinesis_firehose	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
kinesis_firehose	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	0%(500)	✅
kinesis_streams	stdstream	Log Loss	✅	✅	0%(10932)
		Log Duplication	0%(500)	0%(23114)	0%(117546)
kinesis_streams	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	0%(29711)
s3	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
s3	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅

plugin	source		1 MB/s	2 MB/s	3 MB/s
cloudwatch_logs	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
cloudwatch_logs	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅

Note:

• The green check ✅ in the table means no log loss or no log duplication.
• Number in parentheses means the number of records out of total records. For example, 0%(1064/1.8M) under 30Mb/s throughput means 1064 duplicate records out of 18M input records by which log duplication percentage is 0%.
• Log loss is the percentage of data lost and log duplication is the percentage of duplicate logs received at the destination. Your results may differ because they can be influenced by many factors like different configs and environment settings. Log duplication is exclusively caused by partially succeeded batches that were retried, which means it is random.

AWS for Fluent Bit 2.29.0

2.29.0

This release includes:

• Fluent Bit 1.9.10
• Amazon CloudWatch Logs for Fluent Bit 1.9.1
• Amazon Kinesis Streams for Fluent Bit 1.10.1
• Amazon Kinesis Firehose for Fluent Bit 1.7.1

Compared to 2.28.4 this release adds:

• Feature - Add store_dir_limit_size option fluentbit-docs:971
• Feature - New filter for AWS ECS Metadata fluentbit:5898
• Enhancement - Different user agent on windows vs linux fluentbit:6325
• Bug - Resolve Fluent Bit networking hangs affecting CloudWatch plugin by migrating to async networking + sync core scheduler fluentbit:6339

---

We’ve run the new released image in our ECS load testing framework and here is the result. This testing result provides benchmarks of aws-for-fluent-bit under different input load. Learn more about the load test.

plugin	source		20 MB/s	25 MB/s	30 MB/s
kinesis_firehose	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	0%(5870)	✅
kinesis_firehose	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	0%(18460)
kinesis_streams	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
kinesis_streams	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
s3	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
s3	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅

plugin	source		1 MB/s	2 MB/s	3 MB/s
cloudwatch_logs	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
cloudwatch_logs	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅

Note:

• The green check ✅ in the table means no log loss or no log duplication.
• Number in parentheses means the number of records out of total records. For example, 0%(1064/1.8M) under 30Mb/s throughput means 1064 duplicate records out of 18M input records by which log duplication percentage is 0%.
• Log loss is the percentage of data lost and log duplication is the percentage of duplicate logs received at the destination. Your results may differ because they can be influenced by many factors like different configs and environment settings. Log duplication is exclusively caused by partially succeeded batches that were retried, which means it is random.

AWS for Fluent Bit 2.28.4

2.28.4

This release includes:

• Fluent Bit 1.9.9
• Amazon CloudWatch Logs for Fluent Bit 1.9.1
• Amazon Kinesis Streams for Fluent Bit 1.10.1
• Amazon Kinesis Firehose for Fluent Bit 1.7.1

Important Note:

• Two security vulnerabilities were found in amazonlinux which we use as our base image- ALAS-40674 and ALAS-32207. This new image will be based on an updated version of amazonlinux that resolves this CVE.

Compared to 2.28.3 this release adds the following feature that we are working on getting accepted upstream:

• enhancement - Separate AWS User Agents for windows and linux fluentbit:6325

---

We’ve run the new released image in our ECS load testing framework and here is the result. This testing result provides benchmarks of aws-for-fluent-bit under different input load. Learn more about the load test.

plugin	source		20 MB/s	25 MB/s	30 MB/s
kinesis_firehose	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	0%(5490)	✅
kinesis_firehose	tcp	Log Loss	✅	✅	0%(17696)
		Log Duplication	✅	✅	0%(21306)
kinesis_streams	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	0%(53500)	0%(1000)
kinesis_streams	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
cloudwatch_logs	stdstream	Log Loss	16%(99000)	28%(339777)	25%(458092)
		Log Duplication	✅	✅	✅
cloudwatch_logs	tcp	Log Loss	14%(84144)	18%(222962)	43%(779277)
		Log Duplication	✅	✅	✅
s3	stdstream	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅
s3	tcp	Log Loss	✅	✅	✅
		Log Duplication	✅	✅	✅

Note:

• The green check ✅ in the table means no log loss or no log duplication.
• Number in parentheses means the number of records out of total records. For example, 0%(1064/1.8M) under 30Mb/s throughput means 1064 duplicate records out of 18M input records by which log duplication percentage is 0%.
• For CloudWatch output, the only safe throughput where we consistently don't see log loss is 1 MB/s. At 2 Mb/s and beyond, we occasionally see some log loss and throttling.
• Log loss is the percentage of data lost and log duplication is the percentage of duplicate logs received at the destination. Your results may differ because they can be influenced by many factors like different configs and environment settings. Log duplication is exclusively caused by partially succeeded batches that were retried, which means it is random.