Changes

This section includes fixes which change existing functionality. Fixes which don’t involve changes to functionality are documented in Resolved Issues.

If a change relates to a Zendesk issue, the associated Zendesk reference number is displayed, for example (10118).

Changes in NuoDB 8.0

The synchronization of system properties now occurs during garbage collection

Until garbage collection runs, the contents of MEMPROPERTIES system table may differ from the PROPERTIES system table. For more information, see MEMPROPERTIES System Table Description.

Several rate-based metrics now report in per-second units

These rate-based metrics previously used per-ten-second units. For example, a metric that formerly reported 23 events will now report 2 events. For more information, see Metrics Published by Database Processes.

If your application uses these metrics, please re-test and confirm their usage. If necessary, make corrections by relabeling the metric units to "per-second" or by multiplying results by a factor of 10 to revert metrics to "per-ten-second" units. Please contact NuoDB Support, if you have any questions.

The PercentStall metric now reports the non-idle time that a thread spends waiting for another thread to release exclusive access to an object

Earlier, PercentStall reported the active time spent by a thread for another thread to release exclusive access at the end of a stall. As a result, if the wait lasted longer than a monitor reporting period, it could report values greater than 100. For more information, see Metrics Published by Database Processes.

The DATAAFFINITYMAPPING SQL table function now returns PARTITION and STARTID instead of PARTITION and NODEID

For more information, see DATAAFFINITYMAPPING.

Added limits on writes to the following columns

  • STRING and BINARY columns are limited to 1 GB

  • CLOB and BLOB columns are limited to 1.9 GB

For more information, see SQL Binary Types and SQL String and Character Types.

VER is no longer a reserved SQL keyword and can be used as an identifier

For more information, see Reserved SQL Keywords

Changes in Previous NuoDB 7.0.x Releases

Setting the always-recompile-queries connection property to true now fully recompiles only client queries, not all queries

For more information, see Connection Properties.

Removed FOLLOWING, GROUPS, PRECEDING, and UNBOUNDED from the list of reserved keywords

These keywords can now be used for identifiers in SQL statements. For more information, see Reserved SQL Keywords.

Removed support for GRANT CREATE on all system tables, all system views, and all pseudo tables except the SCHEMAS and USERS system tables

For more information, see GRANT.

The CURRENT_TIME and CURRENT_TIMESTAMP functions now support precisions parameter ranging from 0 to 6

For more information, see SQL Date and Time Functions and Operators.

All SQL query results returned in the NuoDB SQL Procedural Language must now be assigned

If a SQL query result returned in the NuoDB SQL Procedural Language was not fully consumed, it may report an error. To avoid this error, make sure that query results are fully consumed using a RETURN clause or by assigning the query results to local variables.

For more information, see SQL Procedural Conditional Statements.

Added FOLLOWING, `GROUPS, PRECEDING, and UNBOUNDED as reserved SQL keywords

For more information, see SQL Keywords.

Renamed the NDV field to APPROX_NDV in GETSCALARSTATS

For more information, see GETSCALARSTATS.

Renamed the SPAN field to HISTOGRAMSPAN in the STATSCACHE system table

For more information, see STATSCACHE System Table Description.

Reserved NORECOMPUTE in the COLUMNGROUPS system table for future use

For more information, see COLUMNGROUPS System Table Description.

Disabled the use of MERGE TREE on temporary tables (14531)

For more information, see MERGE TREE Indexing.

Changed FIELDS, METADATATYPE, and NAME columns to case-insensitive in the following SQL table functions

  • GETAGGREGATEDSCALARSTATS

  • GETHISTOGRAMS

  • GETSCALARSTATS

  • GETSTATSMETADATA

In Kubernetes environments, the timeout for connecting to the Kubernetes API server has been changed to 5 seconds

Memory used by idle statements is now garbage collected to avoid Out-Of-Memory (OOM) failures

The SQL statistics cache is refreshed after the automatic generation of statistics

Rather than ignoring an invalid database option, a TE or SM will refuse to start and report an error