Adds the counters RowsSent and RowsSentRate to the PLAN_ROOT_SINK
section of the profile:
PLAN_ROOT_SINK:
- PeakMemoryUsage: 4.01 MB (4202496)
- RowBatchGetWaitTime: 0.000ns
- RowBatchSendWaitTime: 0.000ns
- RowsSent: 10 (10)
- RowsSentRate: 416.00 /sec
RowsSent tracks the number of rows sent to the PlanRootSink via
PlanRootSink::Send. RowsSentRate tracks the rate that rows are sent to
the PlanRootSink.
Adds the counters NumRowsFetched, NumRowsFetchedFromCache, and
RowMaterializationRate to the ImpalaServer section of the profile.
ImpalaServer:
- ClientFetchWaitTimer: 11.999ms
- NumRowsFetched: 10 (10)
- NumRowsFetchedFromCache: 10 (10)
- RowMaterializationRate: 9.00 /sec
- RowMaterializationTimer: 1s007ms
NumRowsFetched tracks the total number of rows fetched by the query,
but does not include rows fetched from the cache. NumRowsFetchedFromCache
tracks the total number of rows fetched from the query results cache.
RowMaterializationRate tracks the rate at which rows are materialized.
RowMaterializationTimer already existed and tracks how much time is
spent materializing rows.
Testing:
* Added tests to test_fetch_first.py and query_test/test_fetch.py
* Enabled some tests in test_fetch_first.py that were pending
the completion of IMPALA-8819
* Ran core tests
Change-Id: Id9e101e2f3e2bf8324e149c780d35825ceecc036
Reviewed-on: http://gerrit.cloudera.org:8080/14180
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Reviewed-by: Sahil Takiar <stakiar@cloudera.com>
Changes the Coordinator to release admitted memory when each Backend
completes, rather than waiting for the entire query to complete before
releasing admitted memory. When the Coordinator detects that a Backend
has completed (via ControlService::ReportExecStatus) it updates the
state of the Backend in Coordinator::BackendResourceState.
BackendResourceState tracks the state of the admitted resources for
each Backend and decides when the resources for a group of Backend
states should be released. BackendResourceState defines a state machine
to help coordinate the state of the admitted memory for each Backend.
It guarantees that by the time the query is shutdown, all Backends
release their admitted memory.
BackendResourceState implements three rules to control the rate at
which the Coordinator releases admitted memory from the
AdmissionController:
* Resources are released at most once every 1 second, this prevents
short lived queries from causing high load on the admission controller
* Resources are released at most O(log(num_backends)) times; the
BackendResourceStates can release multiple BackendStates from the
AdmissionController at a time
* All pending resources are released if the only remaining Backend is
the Coordinator Backend; this is useful for result spooling where all
Backends may complete, except for the Coordinator Backend
Exposes the following hidden startup flags to help tune the heuristics
above:
--batched_release_decay_factor
* Defaults to 2
* Controls the base value for the O(log(num_backends)) bound when
batching the release of Backends.
--release_backend_states_delay_ms
* Defaults to 1000 milliseconds
* Controls how often Backends can release their resources.
Testing:
* Ran core tests
* Added new tests to test_result_spooling.py and
test_admission_controller.py
Change-Id: I88bb11e0ede7574568020e0277dd8ac8d2586dc9
Reviewed-on: http://gerrit.cloudera.org:8080/14104
Reviewed-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
Tested-by: Impala Public Jenkins <impala-public-jenkins@cloudera.com>
