System information

Resolved Issues in Greenplum Database 4.2.7.3 5
Greenplum Database 4.2.7.3 Release Notes
20466 Catalog and
Metadata
4.2.7.1 If a constraint is created with the same name as an existing constraint,
the only way to remove the second constraint was to drop all
constraints.
The fix is to not allow duplicate constraint names. While creating a new
table or alter table for new constraint creation, duplicate constraint
name are not allowed.
19965 Query
Execution
4.2.7.1 In rare cases, wrong results were returned for a query when during
query processing the result of an operation is shared between multiple
parts of a query through a Shared Scan operator, and the readers of
the Shared Scan appear on both sides of a Join operator, and at least
two of the readers appear in the same plan slice. In this case, one of
the readers might read only zero tuples from the Shared Scan operator.
This has been fixed in this release.
The fix for this issue might increase the memory consumption for
certain queries where these conditions hold.
19930 Monitoring:
gpperfmon
server
4.2.7.1 In some cases, the Greenplum Command Center web server process
could consume a very large amount of memory. This caused the host
machine to run out of memory.
18646 Query
Execution
4.2.7.1 For some SQL queries that generated a query plan with Shared Scan
nodes above a Materialize node, the memory for the Shared Scan
node was being freed twice. In some cases, this caused a SEGV.
18635 Monitoring:
gpperfmon
server
4.2.7.1 In some cases, Greenplum Command Center incorrectly listed SQL
queries as active that were no longer active.
21725 Backup and
Restore
4.2.7 The following gpcrondump utility issues were caused when the root
table of a partitioned table is in a different schema than one of its
partitions (child tables) in a database:
• Creating a backup with the gpcrondump utility and specifying a list of
tables failed.
• Creating a backup of a database with the gpcrondump utility
succeeded. However, the database could not be restored.
21625 Dispatch 4.2.7 It was possible to send a query cancel error message from a QE
process to the QD process for a query that had completed. This caused
multiple statement executions to be out of sync between the QE and
the QD. In some cases, this led to the QE process to be hung
indefinitely.
21522 Backup and
Restore
4.2.7 The Greenplum Database utility pg_dump printed information-level
messages (messages with the label [INFO]) to stderr that were not
printed in previous releases. These messages were printed even when
pg_dump completed without errors.
19476 Client Access
Methods and
Tools
4.2.7 Running multiple gpload sessions simultaneously that loaded data into
the same table resulted in inconsistent data in the table.
19259 Transaction
Management
4.2.7 During high load situations, the clean up of an out-of-memory condition
was not handled correctly and caused a PANIC signal to be issued.
18858 Fault Detection
and
Transitions
4.2.7 Greenplum Database hung indefinitely due a disk that was failing or
had failed. Monitoring for hardware issues has been enhanced in
Greenplum Database.
Table 1 Resolved Issues in 4.2.7.3
Issue
Number
Category
Resolved
in
Description