Where resolution is associated with a reported issue, a Zendesk reference number is displayed, for example (8964).
Resolved in NuoDB 3.2.2
This section provides a summary of issues resolved in the release of NuoDB 3.2.2.
- During write-heavy workloads, queries against tables not accessed in a long time are excessively throttled due to infrequent access of the tables. (8964)
Logic, that excessively throttled queries against infrequently accessed tables, has been corrected.
- In rare scenarios, NuoDB Loader's export utility may fail. This can occur when the encoded representation of the value of a column in .CSV format is larger than the actual value of the column (that is, when some characters in that column need to be escaped). (8936)
This issue has been resolved.
- Where query timeout is set, and the query execution time exceeds the timeout, certain conditions result in a session disconnection instead of query failure. (8989)
These conditions have been resolved so that the query is aborted after the query timeout.
- When using the C driver to connect to NuoDB, the C driver hangs when closing a connection following a network error. (8945)
When an exception is thrown due to a network error, all statements in a connection are now closed and released correctly.
- The JDBC driver returns an incorrect value for COLUMN_SIZE column metadata for some data types. (8895)
COLUMN_SIZE metadata values are now returned correctly.
- If there are brokers in the UNKNOWN state, NuoDB Manager's
show domain health verbose Detailed command fails, displaying a NodeResponsiveness message missing error. (9014)
show domain health verbose Detailed command no longer fails in the scenario described.
- NuoDB Manager's
restart database command may take a long time to return if there is an unreachable broker, even when the
timeout option is set.
When using the
restart database command with the
timeout parameter set, the command now returns after the timeout period if the broker is not reachable.
- Certain situations can result in numerous data atom files that are smaller than 1 KB. (7907)
An error, that caused data atom files less than 1 KB to be created, was corrected.
- NuoDB Manager's
restart database command, with the
incremental true or the
incremental any option specified, fails if any Storage Manager is in the UNREACHABLE state (due to its broker not being reachable). (9007)
restart database command, with
incremental true or
incremental any specified, now attempts to restart the database and ignore any brokers that are not reachable. If Storage Groups are being used and if any Storage Group is not available (because there are no running Storage Managers serving that Storage Group), then the
restart database command returns an error.
(There cannot be any Storage Managers in the UNREACHABLE state if any Storage Group is not available.)
- The Conversion type to “blob” is not implemented. error message is displayed when updating values in a table where a column data type is converted from
VARBINARY and then an index containing that column is added. (9018)
This issue has been resolved.
- Where the
ping-timeout database option is enabled, the starting of a database process may hang if that new process attempts to enter the domain through another process that is no longer available.
If a new process cannot start because it cannot join the domain after 60 seconds, the process exits.
- Unable to deploy NuoDB DataSource in WebSphere Application Server. (8988)
A number of issues which were affecting the ability to deploy a NuoDB DataSource in WebSphere Application Server have been resolved.
Resolved in NuoDB 3.2.1
This section provides a summary of issues resolved in the release of NuoDB 3.2.1.
- If you attempt to create multiple unique indexes on the same columns, an error is now generated. (8686, 8692)
- Unique trigger and index names are now enforced in a single schema.
- Previously, a running hot copy operation prevented the storage manager from being stopped. Now, if a hot copy operation is running when the storage manager is stopped, the hot copy operation is aborted to allow the shutdown to be performed. An error message is also logged. (8816)
- An error is now generated when you insert a (rounded) value that overflows the precision specified for a column.
- You can no longer execute multiple concurrent incremental hot copy operations.
- An error is now generated when you attempt to execute an
ALTER TABLE statement to change the
NOT NULL constraint of a primary key column. (NULL values have never been allowed for primary key columns.) (8687)
- Corrected an error that resulted in transaction engines exiting when
sql-statement-explain-plans logging was enabled. (8774)
- Corrected an error where an INTEGER column used for a partition is compared to a decimal value.
- Corrected an error with joins using an
IN predicate inside an
ON clause joining a partitioned table.
- A condition which resulted in storage managers exiting with the "Archive failure: 'rename' failed with No such file or directory." error has been resolved. (8083)
- Resolved an issue, introduced in NuoDB 3.2, where
UPDATE statements for strings longer than the target column were being truncated. An error message is now returned. (8850)
- Resolved a rare condition where, during Journal Hot Copy, the storage manager exits with a "Journal file ... *.chc not found" error. (8863)
- Previously, automatically generated system constraint names could be longer than 128 characters. This resulted in constraints that could not be dropped. Now, an error is generated if you attempt to create a constraint name longer than 128 characters. This change also allows existing constraints longer than 128 characters to be dropped. Existing constraints longer than 128 characters should be dropped and recreated with names shorter than 128 characters. (8769)
- Query plan selection by the NuoDB Query Optimizer has been improved by taking into account runtime constants such as
Resolved in NuoDB 3.2
This section provides a summary of issues resolved in the release of NuoDB 3.2.
- Index name uniqueness is now enforced when adding constraints using the
CREATE TABLE statement.
- Return data types for user-defined functions are now validated correctly.
- Values are now returned with correct precision when performing arithmetic on literal values with trailing decimal zeros.
- Calculations on numeric types do not always return accurate precision values.
- When RestSvc is disconnected from a broker, the Connection refused error is logged.
- Data types for
DEFAULT values for user-defined functions are now validated correctly.
- When using the
DATE_TO_STR function, inserting a date argument with year value of
0001 produces unexpected results. (8544)
GETUPDATECOUNT function now returns values when used as part of a multiple SQL statement (for SQL statements separated by a semi-colon) operation.
- Unless specified (using
--syslog), error messages are not logging to syslog files.
- Improved plan selection when there are multiple indexes with the same common prefixes on a table. (8477, 8536)
- NuoDB Manager's
log domain and
log database commands continue to run when one or more database processes exit.
GETUPDATECOUNT function now returns the update count, following a
DELETE operation on cursor using the
WHERE CURRENT OF statement.
- Table partitions are not eliminated when a
LIKE predicate on a partitioned column starts with a wildcard.
- Duplicate key errors from triggers are now exposed to users.
- Corrected performance issue related to numeric comparison with parameters. (7828)
- Corrected issues for the JDBC
minIdle connection property.