def _report_sli(self, sli: Dict[str, Any]) -> str:
"""Create report for workflow quality SLI.
@param sli: It's a dict of SLI associated with the SLI type.
"""
html_inputs = self._evaluate_sli(sli=sli)
last_week_data = pd.DataFrame()
if not self.configuration.dry_run:
sli_path = f"{self._SLI_NAME}/{self._SLI_NAME}-{self.configuration.last_week_time}.csv"
last_week_data = retrieve_thoth_sli_from_ceph(
self.configuration.ceph_sli, sli_path, self.total_columns)
for component in self.configuration.registered_workflows:
if not last_week_data.empty:
old_value = last_week_data[component].values[0]
change = evaluate_change(
old_value=old_value,
new_value=html_inputs[component]["value"]) # type: ignore
html_inputs[component]["change"] = change # type: ignore
report = HTMLTemplates.thoth_workflows_quality_template(
html_inputs=html_inputs)
return report
def _evaluate_sli(self, sli: Dict[str, Any]) -> Dict[str, Any]:
"""Evaluate SLI for report for Thoth Integrations Users SLI.
@param sli: It's a dict of SLI associated with the SLI type.
"""
html_inputs: Dict[str, Any] = {}
last_week_data = pd.DataFrame()
if not self.configuration.dry_run:
sli_path = f"{self._SLI_NAME}/{self._SLI_NAME}-{self.configuration.last_week_time}.csv"
last_week_data = retrieve_thoth_sli_from_ceph(self.configuration.ceph_sli, sli_path, self.total_columns)
for thoth_integration in ["KEBECHET"]:
name = thoth_integration.lower()
html_inputs[name] = {}
thoth_integration_counts_total = sli[f"{thoth_integration}_counts_users_total"]
if thoth_integration_counts_total != "ErrorMetricRetrieval":
html_inputs[name]["total_users"] = thoth_integration_counts_total
else:
html_inputs[name]["total_users"] = "N/A"
if not last_week_data.empty and thoth_integration_counts_total != "ErrorMetricRetrieval":
old_value = last_week_data[last_week_data["integration"] == name]["total_users"].values[0]
change = evaluate_change(old_value=old_value, new_value=thoth_integration_counts_total)
html_inputs[name]["new"] = change
else:
html_inputs[name]["new"] = "N/A"
return html_inputs
def _evaluate_sli(self) -> List[Dict[str, Any]]:
"""Evaluate SLI for report for Thoth Adviser Justifications Errors SLI.
@param sli: It's a dict of SLI associated with the SLI type.
"""
html_inputs = []
total_justifications: Dict[str, Any] = {}
days = self.configuration.adviser_inputs_analysis_days
delta = datetime.timedelta(days=1)
if not self.configuration.dry_run:
e_time = self.configuration.start_time.strftime('%Y-%m-%d').split(
"-")
current_end_time = datetime.date(
year=int(e_time[0]),
month=int(e_time[1]),
day=int(e_time[2]),
) + datetime.timedelta(days=1)
current_initial_date = current_end_time - datetime.timedelta(
days=days)
while current_initial_date < current_end_time:
_LOGGER.info(
f"Analyzing for start date: {current_initial_date}")
sli_path = f"{self._SLI_NAME}/{self._SLI_NAME}-{current_initial_date}.csv"
daily_justifications_df = retrieve_thoth_sli_from_ceph(
self.configuration.ceph_sli,
sli_path,
[c for c in self.total_columns if c != "timestamp"],
)
for message in daily_justifications_df["message"].unique():
for adviser_version in daily_justifications_df[
"adviser_version"].unique():
subset_df = daily_justifications_df[
(daily_justifications_df["message"] == message)
& (daily_justifications_df["adviser_version"] ==
adviser_version)]
if subset_df.shape[0] < 1:
continue
counts = subset_df["total"].values[0]
message_type = subset_df["type"].values[0]
if message_type != "ERROR":
continue
if adviser_version not in total_justifications:
total_justifications[adviser_version] = {}
total_justifications[adviser_version][
message] = counts
else:
if message not in total_justifications[
adviser_version]:
total_justifications[adviser_version][
message] = counts
else:
total_justifications[adviser_version][
message] += counts
current_initial_date += delta
for adviser_version, justifications_info in total_justifications.items(
):
total_errors = 0
for _, errors_counts in justifications_info.items():
total_errors += errors_counts
for justification, counts in justifications_info.items():
if not counts:
total = "0"
percentage = 0
else:
total = "+" + "{}".format(int(counts))
percentage = counts / total_errors
html_inputs.append(
{
"adviser_version": adviser_version,
"justification": justification,
"total": total,
"percentage": abs(round(percentage * 100, 3)),
}, )
return html_inputs
def _evaluate_sli(self) -> Dict[str, Any]:
"""Evaluate SLI for report for Thoth Hardware SLI.
@param sli: It's a dict of SLI associated with the SLI type.
"""
html_inputs: Dict[str, Any] = {}
total_quantity: Dict[str, Any] = {}
days = self.configuration.adviser_inputs_analysis_days
delta = datetime.timedelta(days=1)
if not self.configuration.dry_run:
e_time = self.configuration.start_time.strftime('%Y-%m-%d').split("-")
current_end_time = datetime.date(
year=int(e_time[0]), month=int(e_time[1]), day=int(e_time[2]),
) + datetime.timedelta(days=1)
current_initial_date = current_end_time - datetime.timedelta(days=days)
while current_initial_date < current_end_time:
_LOGGER.info(f"Analyzing for start date: {current_initial_date}")
sli_path = f"{self._SLI_NAME}/{self._SLI_NAME}-{current_initial_date}.csv"
daily_quantity_df = retrieve_thoth_sli_from_ceph(
self.configuration.ceph_sli,
sli_path,
[c for c in self.total_columns if c != "timestamp"],
)
for cpu_model in daily_quantity_df["cpu_model"].unique():
subset_df = daily_quantity_df[daily_quantity_df["cpu_model"] == cpu_model]
hardware = f"{subset_df['cpu_model'].values[0]}-{subset_df['cpu_family'].values[0]}"
if hardware not in total_quantity:
total_quantity[hardware] = {
"cpu_model": subset_df['cpu_model'].values[0],
"cpu_family": subset_df['cpu_family'].values[0],
"counts": subset_df["total"].values[0],
}
else:
total_quantity[hardware]["counts"] += subset_df["total"].values[0]
current_initial_date += delta
total_ = 0
for _, total_counts in total_quantity.items():
total_ += total_counts["counts"]
for hardware_, hardware_info in total_quantity.items():
html_inputs[hardware_] = {}
html_inputs[hardware_]["cpu_model"] = int(hardware_info["cpu_model"])
html_inputs[hardware_]["cpu_family"] = int(hardware_info["cpu_family"])
if not hardware_info["counts"]:
html_inputs[hardware_]["new"] = 0
html_inputs[hardware_]["percentage"] = 0
else:
percentage = hardware_info["counts"] / total_
html_inputs[hardware_]["new"] = "+" + "{}".format(int(hardware_info["counts"]))
html_inputs[hardware_]["percentage"] = abs(round(percentage * 100, 3))
return html_inputs
def _evaluate_sli(self) -> List[Dict[str, Any]]:
"""Evaluate SLI for report for Thoth Adviser Reports Statistics SLI.
@param sli: It's a dict of SLI associated with the SLI type.
"""
html_inputs = []
total_statistics: Dict[str, Any] = {}
days = self.configuration.adviser_inputs_analysis_days
delta = datetime.timedelta(days=1)
if not self.configuration.dry_run:
e_time = self.configuration.start_time.strftime('%Y-%m-%d').split(
"-")
current_end_time = datetime.date(
year=int(e_time[0]),
month=int(e_time[1]),
day=int(e_time[2]),
) + datetime.timedelta(days=1)
current_initial_date = current_end_time - datetime.timedelta(
days=days)
while current_initial_date < current_end_time:
_LOGGER.info(
f"Analyzing for start date: {current_initial_date}")
sli_path = f"{self._SLI_NAME}/{self._SLI_NAME}-{current_initial_date}.csv"
daily_statistics_df = retrieve_thoth_sli_from_ceph(
self.configuration.ceph_sli,
sli_path,
[
c for c in self.total_columns
if c not in ["timestamp", "datetime"]
],
)
for adviser_version in daily_statistics_df[
"adviser_version"].unique():
subset_df = daily_statistics_df[
daily_statistics_df["adviser_version"] ==
adviser_version]
s_counts = 0
f_counts = 0
if not subset_df.empty:
s_counts = subset_df[
subset_df["adviser_version"] ==
adviser_version]["success"].values[0]
f_counts = subset_df[
subset_df["adviser_version"] ==
adviser_version]["failure"].values[0]
if adviser_version not in total_statistics:
total_statistics[adviser_version] = {}
total_statistics[adviser_version]["success"] = s_counts
total_statistics[adviser_version]["failure"] = f_counts
else:
total_statistics[adviser_version][
"success"] += s_counts
total_statistics[adviser_version][
"failure"] += f_counts
current_initial_date += delta
for adviser_version, statistics_info in total_statistics.items():
total = statistics_info["success"] + statistics_info["failure"]
success_p = statistics_info["success"] / total
failure_p = statistics_info["failure"] / total
html_inputs.append(
{
"adviser_version": adviser_version,
"success_p": abs(round(success_p * 100, 3)),
"failure_p": abs(round(failure_p * 100, 3)),
}, )
return html_inputs