AWS dms: DMS release notes: bug-fix entries added and reworded
Summary
Adds and rewrites DMS release-note entries describing bug fixes across sources/targets (DocumentDB/MongoDB CDC log noise, S3 target empty directory cleanup, data validation crash, S3 Parquet memory growth, MongoDB socketTimeoutMS attribute, SQL Server CDC data handling, Db2 validation, PostgreSQL SSL connection crash, Redshift and Redshift Zero-ETL full-load crashes), and corrects the wide-character/Oracle wording.
Security assessment
The changed lines are release-note descriptions of functional/reliability defects (memory growth, task crashes, incorrect character/binary values, log noise). Even the closest-to-security entry, the PostgreSQL SSL connection crash fix, addresses availability/stability of TLS connections rather than a vulnerability, credential leak, or authz flaw; no CVE or advisory is referenced, and no new security guidance or configuration is documented.
Evidence
+PostgreSQL SSL connection crash issue | Fixed an issue for PostgreSQL endpoints on DMS 3.5.4 replication instances where the replication process crashed and restarted when the endpoint used an SSL connection. SSL connections are now stable.
Diff
diff --git a/dms/latest/userguide/CHAP_ReleaseNotes.md b/dms/latest/userguide/CHAP_ReleaseNotes.md index 2fbfdf2..d9e8c00 100644 --- a//dms/latest/userguide/CHAP_ReleaseNotes.md +++ b//dms/latest/userguide/CHAP_ReleaseNotes.md @@ -137 +137 @@ Amazon DocumentDB and MongoDB change stream crash issue | Fixed an issue for Ama -Wide character data handling issue | Fixed an issue where wide character (WSTR) values were not built correctly into SQL literals, which wrote incorrect character data to the target. +Wide character data handling issue | Fixed an issue for Oracle targets that use a non-Unicode database character set, where wide character (WSTR) values were not built correctly into SQL literals, which wrote incorrect character data to the target. @@ -141,0 +142,7 @@ Oracle partition query ordering update | Added deterministic ordering to the Ora +Amazon DocumentDB and MongoDB task log noise issue | Fixed an issue for Amazon DocumentDB and MongoDB sources where change data capture (CDC) wrote excessive warning messages to the task log, which made real errors harder to find. The task log is now quieter during normal operation. +Amazon S3 target empty directory cleanup issue | Fixed an issue for Amazon S3 targets where empty date-partitioned directories were left on the replication instance during change data capture (CDC) and consumed storage over time. Empty directories are now removed automatically. +Data validation crash issue | Fixed an issue where data validation stopped the task with a crash, observed on memory-constrained replication instances. Data validation now completes without crashing in this case. +Amazon S3 Parquet source memory issue | Fixed issues for Amazon S3 sources that use Parquet where memory use grew during ongoing replication until the replication instance ran out of memory. Memory use is now stable. +MongoDB socketTimeoutMS connection attribute issue | Fixed an issue for MongoDB sources where the socketTimeoutMS extra connection attribute was ignored when entered with the uppercase spelling shown in the MongoDB documentation, so the socket timeout stayed at the default of 300 seconds. The documented spelling is now applied. +Microsoft SQL Server CDC data handling issue | Fixed an issue for Microsoft SQL Server sources where change data capture (CDC) wrote incorrect character or binary values for certain columns, or stopped the task for tables that have no primary key or unique index. These tables now replicate correctly. +IBM Db2 source validation issue | Fixed an issue for IBM Db2 sources where data validation failed on DECIMAL primary key columns that contain values with trailing zeros. These columns now validate correctly. @@ -255 +262 @@ Amazon S3 source storage growth issue | Fixed an issue for Amazon S3 sources whe -Amazon S3 Parquet reader memory issue | Fixed an issue for Amazon S3 sources where the Parquet reader accumulated memory over time. +Amazon S3 Parquet source memory issue | Fixed issues for Amazon S3 sources that use Parquet where memory use grew during ongoing replication until the replication instance ran out of memory. Memory use is now stable. @@ -291 +297,0 @@ Amazon DocumentDB and MongoDB change stream crash issue | Fixed an issue for Ama -Wide character data handling issue | Fixed an issue where wide character (WSTR) values were not built correctly into SQL literals, which wrote incorrect character data to the target. @@ -294,0 +301,10 @@ Amazon Redshift load timeout default update | Increased the default Amazon Redsh +Amazon DocumentDB and MongoDB task log noise issue | Fixed an issue for Amazon DocumentDB and MongoDB sources where change data capture (CDC) wrote excessive warning messages to the task log, which made real errors harder to find. The task log is now quieter during normal operation. +Amazon S3 target empty directory cleanup issue | Fixed an issue for Amazon S3 targets where empty date-partitioned directories were left on the replication instance during change data capture (CDC) and consumed storage over time. Empty directories are now removed automatically. +Data validation crash issue | Fixed an issue where data validation stopped the task with a crash, observed on memory-constrained replication instances. Data validation now completes without crashing in this case. +MongoDB socketTimeoutMS connection attribute issue | Fixed an issue for MongoDB sources where the socketTimeoutMS extra connection attribute was ignored when entered with the uppercase spelling shown in the MongoDB documentation, so the socket timeout stayed at the default of 300 seconds. The documented spelling is now applied. +Microsoft SQL Server CDC data handling issue | Fixed an issue for Microsoft SQL Server sources where change data capture (CDC) wrote incorrect character or binary values for certain columns, or stopped the task for tables that have no primary key or unique index. These tables now replicate correctly. +IBM Db2 source validation issue | Fixed an issue for IBM Db2 sources where data validation failed on DECIMAL primary key columns that contain values with trailing zeros. These columns now validate correctly. +PostgreSQL SSL connection crash issue | Fixed an issue for PostgreSQL endpoints on DMS 3.5.4 replication instances where the replication process crashed and restarted when the endpoint used an SSL connection. SSL connections are now stable. +Amazon DocumentDB empty field name issue | Fixed an issue for Amazon DocumentDB targets where a migration stopped with an Empty field name error when source rows contained null or empty string values. These values are now applied correctly. +Amazon Redshift full-load crash issue | Fixed an issue for Amazon Redshift targets on DMS 3.5.4 where a full-load task stopped with a crash. Full-load tasks now complete. +Amazon Redshift Zero-ETL full-load crash issue | Fixed an issue for Amazon Redshift Zero-ETL targets on DMS 3.5.4 where a full-load task stopped with a crash when a table produced more than one data file.