timeWrongJoinImpls += operator['avg_time']
elif isWrongJoinLeftRight(operator, leftChild, rightChild):
numWrongJoinLeftRights += 1
timeWrongJoinLeftRights += operator['avg_time']
else:
numCorrectJoins += 1
timeCorrectJoins += operator['avg_time']
plots.stacked_bar(
[
numCorrectJoins,
numWrongJoinImpls,
numWrongJoinLeftRights,
],
'Number of Joins',
[
'Correct Join %s' % numCorrectJoins,
'Wrong Join Impl %s' % numWrongJoinImpls,
'Wrong Join Left Right %s' % numWrongJoinLeftRights,
],
'Join Correctness',
'stacked_num_join_correctness.png'
)
plots.stacked_bar(
[
timeCorrectJoins / 1000000,
timeWrongJoinImpls / 1000000,
timeWrongJoinLeftRights / 1000000,
],
'Time of Joins',
"_id": "Fragment Start",
"sum_time": query["fragment_start_time"],
"time_pct": query["fragment_start_time"] / sumTimeAllOperators,
}
)
operators.append({"_id": "CodeGen", "sum_time": codeGenTime, "time_pct": codeGenTime / sumTimeAllOperators})
operators.append(
{"_id": "HdfsTableSink", "sum_time": hdfsTableSinkTime, "time_pct": hdfsTableSinkTime / sumTimeAllOperators}
)
if not args.summary:
operators.sort(key=lambda operator: operator["sum_time"], reverse=True)
plots.stacked_bar(
[operator["sum_time"] / 1000000 for operator in operators],
"Time",
["%s %sms" % (operator["_id"], operator["sum_time"] / 1000000) for operator in operators],
"Operator Sum Time (ms)",
"%s_stacked_time.png" % query["_id"],
)
for operator in operators:
if operator["_id"] not in timePctPerOperator:
timePctPerOperator[operator["_id"]] = []
timePctPerOperator[operator["_id"]].append(operator["time_pct"])
if operator["_id"] not in sumTimePerOperator:
sumTimePerOperator[operator["_id"]] = []
sumTimePerOperator[operator["_id"]].append(operator["sum_time"])
numJoins.append(db.operators.find({"query_id": query["_id"], "name": {"$in": ["HASH JOIN", "CROSS JOIN"]}}).count())
