Greenplum® Command Center parameters

This section describes the session and query parameters available in Greenplum® Command Center.

SessionSession

Resources:

• CPU time: Time used by CPU resources to process queries.

• Peak memory: Maximum memory used to process queries.

• Spill: Total size of the files generated and spilled to disk while executing the query.

• Disk R: Number of bytes read from disk while executing the query.

• Disk W: Number of bytes written to disk while executing the query.

• Packets recv: Number of packets received via interconnect.

• Packets sent: Number of packets sent via interconnect.

Session info:

• User: Username.

• Database: DB name.

• Status changed: Date and time of the most recent status change.

• Start time: Session start date and time.

• Application: Connected client’s app name.

• Client host: Connected client’s host name.

• Process ID: Current session ID.

• Client address: Connected client’s IP address.

• Client port: Number of the port the client uses for the connection.

• Transaction start time: Current transaction’s start date and time.

• Waiting: This parameter defines whether the session waits for blocking events to complete.

• Transaction horizon: Earliest transaction ID.

• OID resource group: Resource group object ID.

• Resource group: Current resource group name.

Queries:

• Key of running query: Query key.

• Query text: Text of the query being executed.

• Start time: Date and time when the query started executing.

• Query state: Status of the query being executed.

• Plan ID: Query execution plan ID.

Other resources:

• CPU user time: Time used by CPU resources to run program instructions.

• CPU kernel time: Time used by CPU resources to process system calls.

• Resident set size: Memory used to execute session queries, in bytes.

• Virtual memory size: Memory allocated to the session.

• Chars read: Number of read characters.

• Chars written: Number of written characters.

• Read syscalls: Number of read operations.

• Write syscalls: Number of write operations.

• Cancelled write bytes: Size of data dropped due to a canceled write.

• Packets sent: Number of packets sent via interconnect.

• Packets recv: Number of packets received via interconnect.

Number of rows and optimizer rating:

• Cycles count: Number of plan node calls.

• First tuple time: Time spent processing the first row, in seconds.

• Startup time: Time spent starting the operator, in seconds.

• Total time: Total time spent processing rows.

Number of buffer accesses and other DB counters:

• Shared buffer hits: Number of shared memory blocks fetched from the buffer.

• Shared disk blocks read: Number of shared memory blocks read from disk as they were missing in the buffer.

• Shared blocks dirtied: Number of shared memory blocks that were modified and now require writing to disk.

• Shared disk blocks written: Number of shared memory blocks written to disk.

• Local buffer hits: Number of local blocks fetched from the buffer.

• Local disk blocks read: Number of local blocks read from disk as they were missing in the buffer.

• Local disk blocks written: Number of local blocks written to disk.

• Temp blocks read: Number of temporary file blocks read from disk.

• Temp blocks written: Number of temporary file blocks written to disk.

• Time spent reading blocks: Time spent reading blocks.

• Time spent writing blocks: Time spent writing blocks.

• Spill file count: Number of spill files.

• Planning and resource waiting time: Time spent waiting for resources and planning.

QueryQuery

Resources:

• CPU time: Time used by CPU resources to process queries.

• Peak memory: Maximum memory used to process queries.

• Spill: Total size of the files generated and spilled to disk while executing the query.

• Disk R: Number of bytes read from disk while executing the query.

• Disk W: Number of bytes written to disk while executing the query.

• Packets recv: Number of packets received via interconnect.

• Packets sent: Number of packets sent via interconnect.

• Skew: Data skew (non-uniform distribution across segments).

Query info:

• Query key: Unique attribute assigned to the query during its execution.

• Start time: Query execution start date and time.

• End time: Query execution end date and time.

• Optimizer: Query optimizer.

• Min time: Minimum query execution time.

• Max time: Maximum query execution time.

• User: Username.

• Database: DB name.

• Query ID: Non-unique hash generated from the query text.

Other resources:

• CPU user time: Time used by CPU resources to run program instructions.

• CPU kernel time: Time used by CPU resources to process system calls.

• Resident Set Size: Memory used to execute the query, in bytes.

• Virtual Memory Size: Memory allocated to execute the query.

• Chars read: Number of read characters.

• Chars written: Number of written characters.

• Read syscalls: Number of read operations.

• Write syscalls: Number of write operations.

• Cancelled write bytes: Size of data dropped due to a canceled write.

• Packets sent: Number of packets sent via interconnect.

• Packets recv: Number of packets received via interconnect.

Number of rows and optimizer rating:

• Cycles count: Number of plan node calls.

• First tuple time: Time spent processing the first row, in seconds.

• Startup time: Time spent starting the operator, in seconds.

• Total time: Total time spent processing rows.

Number of buffer accesses and other DB counters:

• Shared buffer hits: Number of shared memory blocks fetched from the buffer.

• Shared disk blocks read: Number of shared memory blocks read from disk as they were missing in the buffer.

• Shared blocks dirtied: Number of shared memory blocks that were modified and now require writing to disk.

• Shared disk blocks written: Number of shared memory blocks written to disk.

• Local buffer hits: Number of local blocks fetched from the buffer.

• Local disk blocks read: Number of local blocks read from disk as they were missing in the buffer.

• Local blocks dirtied: Number of local blocks that were modified and now require writing to disk.

• Local disk blocks written: Number of local blocks written to disk.

• Temp blocks read: Number of temporary file blocks read from disk.

• Temp blocks written: Number of temporary file blocks written to disk.

• Time spent reading blocks: Time spent reading blocks.

• Time spent writing blocks: Time spent writing blocks.

• Spill file count: Number of spill files.

• Planning and resource waiting time: Time spent waiting for resources and planning.

Interconnect network metricsInterconnect network metrics

• Interconnect active connections num: Maximum number of active outbound connections registered.

• Interconnect buffer counting time: Monotonically increasing counter of buffer availability checks before sending packets. It increases with each successful outbound connection.

• Interconnect buffer counting time: Monotonically increasing counter of outbound queue size checks. It increases with every processed connection.

• Interconnect CRC errors: Monotonically increasing counter of packets that failed checksum validation. It applies if gp_interconnect_full_crc is enabled.

• Interconnect disordered pkt num: Monotonically increasing counter of out-of-order packets. It increases if at least one of these conditions is met:

  • The sequence number of the inbound packet does not match the expected value.
  • The inbound packet queue overflowed and the packet was dropped.

• Interconnect duplicated pkt num: Monotonically increasing counter of duplicate packets. It increases when the sequence number of the inbound packet is lower than the highest sequence number of processed packets.

• Interconnect mismatch num: Monotonically increasing counter of mismatched packets. These packets pertain to the previous query or the current one but are not yet ready for processing.

• Interconnect recv ack num: Monotonically increasing counter of acknowledgments for inbound packets.

• Interconnect recv pkt num: Monotonically increasing counter of processed inbound packets.

• Interconnect recv queue size counting time: Monotonically increasing counter of inbound queue size checks. It increases with every processed connection.

• Interconnect retransmits: Monotonically increasing counter of retransmitted packets. Retransmission applies to out-of-order and outdated packets.

• Interconnect snd pkt num: Monotonically increasing counter of sent packets.

• Interconnect startup cached pkt num: Monotonically increasing counter of cached packets received but not yet ready for processing.

• Interconnect status query msg num: Monotonically increasing counter of transmitted status packets. These packets are used to check the connection health if no packet transmission acknowledgments are received for a while.

• Interconnect total buffers: Minimum number of available buffers registered before sending packets.

• Interconnect total capacity: Maximum size of the outbound queue, in packets.

• Interconnect total recv queue size: Maximum size of the inbound queue, in packets.

Learn more about interconnect here.

Greenplum® and Greenplum Database® are registered trademarks or trademarks of Broadcom Inc. in the United States and/or other countries.