iterators_*_rollup

The iterators_*_rollup set of views aggregate the metrics stored in the iterators_* tables. A query iterator refers to a node or operation in a query plan. For example, a sequential scan operation may be one type of iterator in a particular query plan. For each iterator in a query plan, the iterators_* tables store the metrics collected from each segment instance. The iterators_*_rollup views summarize the query iterator metrics across all segments in the system.

The tmid, ssid and ccnt columns are the composite key that uniquely identifies a particular query.

There are three iterators rollup views, all having the same columns:

The iterators_now_rollup view shows iterator data from the interators_now table aggregated across all segments in the system.
The iterators_tail_rollup view shows iterator data from the interators_tail table aggregated across all segments in the system.
The iterators_history_rollup shows iterator data from the interators_history table aggregated across all segments in the system.

See also the iterators_* tables for more information about the query plan iterator types and the metrics collected for each iterator.

Column	Type	Description
`sample_time`	timestamp	The `ctime` from the associated `iterators_*` table.
`tmid`	int	A time identifier for a particular query. All iterator records associated with the query will have the same `tmid`.
`ssid`	int	The session id as shown by the `gp_session_id` parameter. All iterator records associated with the query will have the same `ssid`.
`ccnt`	int	The command number within this session as shown by `gp_command_count` parameter. All iterator records associated with the query will have the same `ccnt`.
`nid`	int	The ID of this query plan node from the slice plan.
`pnid`	int	The `pnid` (slice plan parent node ID) from the associated `iterators_*` table.
`ntype`	text	The `ntype` (node/iterator type) from the associated `iterators_*` table.
`nstatus`	text	The accumulated status of this iterator. Possible values are: Initialize, Executing, or Finished.
`tstart`	timestamp	The average start time for this iterator.
`tduration`	numeric	The average execution time for this iterator.
`pmemsize`	numeric	The average work memory allocated by the Greenplum planner to this iterator’s query processes.
`pmemmax`	numeric	The average of the maximum planner work memory used by this iterator’s query processes.
`memsize`	numeric	The average OS memory allocated to this iterator’s processes.
`memresid`	numeric	The average resident memory allocated to this iterator’s processes (as opposed to shared memory).
`memshare`	numeric	The average shared memory allocated to this iterator’s processes.
`cpu_elapsed`	numeric	Sum of the CPU usage of all segment processes executing this iterator.
`cpu_currpct`	double precision	The current average percentage of CPU utilization used by this iterator’s processes. This value is always zero for historical (completed) iterators.
`rows_out`	numeric	The total number of actual rows output for this iterator on all segments.
`rows_out_est`	numeric	The total number of output rows for all segments as estimated by the query planner.
`skew_cpu`	numeric	Coefficient of variation for `cpu_elapsed` of iterators across all segments for this query, multiplied by 100. For example, a value of .95 is rendered as 95.
`skew_rows`	numeric	Coefficient of variation for `rows_out` of iterators across all segments for this query, multiplied by 100. For example, a value of .95 is rendered as 95.
`m0`	text	The name (`m0_name`), unit of measure (`m0_unit)`, average actual value (`m0_val`), and average estimated value (`m0_est`) for this iterator metric across all segments. The `m0` metric is always rows for all iterator types.
`m1`	text	The name (`m1_name`), unit of measure (`m1_unit`), average actual value (`m1_val`), and average estimated value (`m1_est`) for this iterator metric across all segments.
`m2`	text	The name (`m2_name`), unit of measure (`m2_unit`), average actual value (`m2_val`), and average estimated value (`m2_est`) for this iterator metric across all segments.
`m3`	text	The name (`m3_name`), unit of measure (`m3_unit`), average actual value (`m3_val`), and average estimated value (`m3_est`) for this iterator metric across all segments.
`m4`	text	The name (`m4_name`), unit of measure (`m4_unit`), average actual value (`m4_val`), and average estimated value (`m4_est`) for this iterator metric across all segments.
`m5`	text	The name (`m5_name`), unit of measure (`m5_unit`), average actual value (`m5_val`), and average estimated value (`m5_est`) for this iterator metric across all segments.
`m6`	text	The name (`m6_name`), unit of measure (`m6_unit`), average actual value (`m6_val`), and average estimated value (`m6_est`) for this iterator metric across all segments.
`m7`	text	The name (`m7_name`), unit of measure (`m7_unit`), average actual value (`m7_val`), and average estimated value (`m7_est`) for this iterator metric across all segments.
`m8`	text	The name (`m8_name`), unit of measure (`m8_unit`), average actual value (`m8_val`), and average estimated value (`m8_est`) for this iterator metric across all segments.
`m9`	text	The name (`m9_name`), unit of measure (`m9_unit`), average actual value (`m9_val`), and average estimated value (`m9_est`) for this iterator metric across all segments.
`m10 - m5`	text	Metrics `m10` through `m15` are not currently used by any iterator types.
`t0`	text	The name of the relation (`t0_val`) being scanned by this iterator. This metric is collected only for iterators that perform scan operations such as a sequential scan or function scan.