Coverage for python/lsst/ctrl/bps/htcondor/report_utils.py: 7%
290 statements
« prev ^ index » next coverage.py v7.10.6, created at 2025-10-11 01:37 -0700
1# This file is part of ctrl_bps_htcondor.
2#
3# Developed for the LSST Data Management System.
4# This product includes software developed by the LSST Project
5# (https://www.lsst.org).
6# See the COPYRIGHT file at the top-level directory of this distribution
7# for details of code ownership.
8#
9# This software is dual licensed under the GNU General Public License and also
10# under a 3-clause BSD license. Recipients may choose which of these licenses
11# to use; please see the files gpl-3.0.txt and/or bsd_license.txt,
12# respectively. If you choose the GPL option then the following text applies
13# (but note that there is still no warranty even if you opt for BSD instead):
14#
15# This program is free software: you can redistribute it and/or modify
16# it under the terms of the GNU General Public License as published by
17# the Free Software Foundation, either version 3 of the License, or
18# (at your option) any later version.
19#
20# This program is distributed in the hope that it will be useful,
21# but WITHOUT ANY WARRANTY; without even the implied warranty of
22# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
23# GNU General Public License for more details.
24#
25# You should have received a copy of the GNU General Public License
26# along with this program. If not, see <https://www.gnu.org/licenses/>.
28"""Utility functions used for reporting."""
30import logging
31import os
32import re
33from pathlib import Path
34from typing import Any
36import htcondor
38from lsst.ctrl.bps import (
39 WmsJobReport,
40 WmsRunReport,
41 WmsSpecificInfo,
42 WmsStates,
43)
45from .common_utils import _htc_status_to_wms_state
46from .lssthtc import (
47 MISSING_ID,
48 WmsNodeType,
49 condor_search,
50 htc_check_dagman_output,
51 htc_tweak_log_info,
52 pegasus_name_to_label,
53 read_dag_info,
54 read_dag_log,
55 read_dag_status,
56 read_node_status,
57 summarize_dag,
58)
60_LOG = logging.getLogger(__name__)
63def _get_status_from_id(
64 wms_workflow_id: str, hist: float, schedds: dict[str, htcondor.Schedd]
65) -> tuple[WmsStates, str]:
66 """Gather run information using workflow id.
68 Parameters
69 ----------
70 wms_workflow_id : `str`
71 Limit to specific run based on id.
72 hist : `float`
73 Limit history search to this many days.
74 schedds : `dict` [ `str`, `htcondor.Schedd` ]
75 HTCondor schedulers which to query for job information. If empty
76 dictionary, all queries will be run against the local scheduler only.
78 Returns
79 -------
80 state : `lsst.ctrl.bps.WmsStates`
81 Status for the corresponding run.
82 message : `str`
83 Message with extra error information.
84 """
85 _LOG.debug("_get_status_from_id: id=%s, hist=%s, schedds=%s", wms_workflow_id, hist, schedds)
87 message = ""
89 # Collect information about the job by querying HTCondor schedd and
90 # HTCondor history.
91 schedd_dag_info = _get_info_from_schedd(wms_workflow_id, hist, schedds)
92 if len(schedd_dag_info) == 1:
93 schedd_name = next(iter(schedd_dag_info))
94 dag_id = next(iter(schedd_dag_info[schedd_name]))
95 dag_ad = schedd_dag_info[schedd_name][dag_id]
96 state = _htc_status_to_wms_state(dag_ad)
97 else:
98 state = WmsStates.UNKNOWN
99 message = f"DAGMan job {wms_workflow_id} not found in queue or history. Check id or try path."
100 return state, message
103def _get_status_from_path(wms_path: str | os.PathLike) -> tuple[WmsStates, str]:
104 """Gather run status from a given run directory.
106 Parameters
107 ----------
108 wms_path : `str` | `os.PathLike`
109 The directory containing the submit side files (e.g., HTCondor files).
111 Returns
112 -------
113 state : `lsst.ctrl.bps.WmsStates`
114 Status for the run.
115 message : `str`
116 Message to be printed.
117 """
118 wms_path = Path(wms_path).resolve()
119 message = ""
120 try:
121 wms_workflow_id, dag_ad = read_dag_log(wms_path)
122 except FileNotFoundError:
123 wms_workflow_id = MISSING_ID
124 message = f"DAGMan log not found in {wms_path}. Check path."
126 if wms_workflow_id == MISSING_ID:
127 state = WmsStates.UNKNOWN
128 else:
129 state = _htc_status_to_wms_state(dag_ad[wms_workflow_id])
131 return state, message
134def _report_from_path(wms_path):
135 """Gather run information from a given run directory.
137 Parameters
138 ----------
139 wms_path : `str`
140 The directory containing the submit side files (e.g., HTCondor files).
142 Returns
143 -------
144 run_reports : `dict` [`str`, `lsst.ctrl.bps.WmsRunReport`]
145 Run information for the detailed report. The key is the HTCondor id
146 and the value is a collection of report information for that run.
147 message : `str`
148 Message to be printed with the summary report.
149 """
150 wms_workflow_id, jobs, message = _get_info_from_path(wms_path)
151 if wms_workflow_id == MISSING_ID:
152 run_reports = {}
153 else:
154 run_reports = _create_detailed_report_from_jobs(wms_workflow_id, jobs)
155 return run_reports, message
158def _report_from_id(wms_workflow_id, hist, schedds=None):
159 """Gather run information using workflow id.
161 Parameters
162 ----------
163 wms_workflow_id : `str`
164 Limit to specific run based on id.
165 hist : `float`
166 Limit history search to this many days.
167 schedds : `dict` [ `str`, `htcondor.Schedd` ], optional
168 HTCondor schedulers which to query for job information. If None
169 (default), all queries will be run against the local scheduler only.
171 Returns
172 -------
173 run_reports : `dict` [`str`, `lsst.ctrl.bps.WmsRunReport`]
174 Run information for the detailed report. The key is the HTCondor id
175 and the value is a collection of report information for that run.
176 message : `str`
177 Message to be printed with the summary report.
178 """
179 messages = []
181 # Collect information about the job by querying HTCondor schedd and
182 # HTCondor history.
183 schedd_dag_info = _get_info_from_schedd(wms_workflow_id, hist, schedds)
184 if len(schedd_dag_info) == 1:
185 # Extract the DAG info without altering the results of the query.
186 schedd_name = next(iter(schedd_dag_info))
187 dag_id = next(iter(schedd_dag_info[schedd_name]))
188 dag_ad = schedd_dag_info[schedd_name][dag_id]
190 # If the provided workflow id does not correspond to the one extracted
191 # from the DAGMan log file in the submit directory, rerun the query
192 # with the id found in the file.
193 #
194 # This is to cover the situation in which the user provided the old job
195 # id of a restarted run.
196 try:
197 path_dag_id, _ = read_dag_log(dag_ad["Iwd"])
198 except FileNotFoundError as exc:
199 # At the moment missing DAGMan log is pretty much a fatal error.
200 # So empty the DAG info to finish early (see the if statement
201 # below).
202 schedd_dag_info.clear()
203 messages.append(f"Cannot create the report for '{dag_id}': {exc}")
204 else:
205 if path_dag_id != dag_id:
206 schedd_dag_info = _get_info_from_schedd(path_dag_id, hist, schedds)
207 messages.append(
208 f"WARNING: Found newer workflow executions in same submit directory as id '{dag_id}'. "
209 "This normally occurs when a run is restarted. The report shown is for the most "
210 f"recent status with run id '{path_dag_id}'"
211 )
213 if len(schedd_dag_info) == 0:
214 run_reports = {}
215 elif len(schedd_dag_info) == 1:
216 _, dag_info = schedd_dag_info.popitem()
217 dag_id, dag_ad = dag_info.popitem()
219 # Create a mapping between jobs and their classads. The keys will
220 # be of format 'ClusterId.ProcId'.
221 job_info = {dag_id: dag_ad}
223 # Find jobs (nodes) belonging to that DAGMan job.
224 job_constraint = f"DAGManJobId == {int(float(dag_id))}"
225 schedd_job_info = condor_search(constraint=job_constraint, hist=hist, schedds=schedds)
226 if schedd_job_info:
227 _, node_info = schedd_job_info.popitem()
228 job_info.update(node_info)
230 # Collect additional pieces of information about jobs using HTCondor
231 # files in the submission directory.
232 _, path_jobs, message = _get_info_from_path(dag_ad["Iwd"])
233 _update_jobs(job_info, path_jobs)
234 if message:
235 messages.append(message)
236 run_reports = _create_detailed_report_from_jobs(dag_id, job_info)
237 else:
238 ids = [ad["GlobalJobId"] for dag_info in schedd_dag_info.values() for ad in dag_info.values()]
239 message = (
240 f"More than one job matches id '{wms_workflow_id}', "
241 f"their global ids are: {', '.join(ids)}. Rerun with one of the global ids"
242 )
243 messages.append(message)
244 run_reports = {}
246 message = "\n".join(messages)
247 return run_reports, message
250def _get_info_from_schedd(
251 wms_workflow_id: str, hist: float, schedds: dict[str, htcondor.Schedd]
252) -> dict[str, dict[str, dict[str, Any]]]:
253 """Gather run information from HTCondor.
255 Parameters
256 ----------
257 wms_workflow_id : `str`
258 Limit to specific run based on id.
259 hist : `float`
260 Limit history search to this many days.
261 schedds : `dict` [ `str`, `htcondor.Schedd` ]
262 HTCondor schedulers which to query for job information. If empty
263 dictionary, all queries will be run against the local scheduler only.
265 Returns
266 -------
267 schedd_dag_info : `dict` [`str`, `dict` [`str`, `dict` [`str` Any]]]
268 Information about jobs satisfying the search criteria where for each
269 Scheduler, local HTCondor job ids are mapped to their respective
270 classads.
271 """
272 _LOG.debug("_get_info_from_schedd: id=%s, hist=%s, schedds=%s", wms_workflow_id, hist, schedds)
274 dag_constraint = 'regexp("dagman$", Cmd)'
275 try:
276 cluster_id = int(float(wms_workflow_id))
277 except ValueError:
278 dag_constraint += f' && GlobalJobId == "{wms_workflow_id}"'
279 else:
280 dag_constraint += f" && ClusterId == {cluster_id}"
282 # With the current implementation of the condor_* functions the query
283 # will always return only one match per Scheduler.
284 #
285 # Even in the highly unlikely situation where HTCondor history (which
286 # condor_search queries too) is long enough to have jobs from before
287 # the cluster ids were rolled over (and as a result there is more then
288 # one job with the same cluster id) they will not show up in
289 # the results.
290 schedd_dag_info = condor_search(constraint=dag_constraint, hist=hist, schedds=schedds)
291 return schedd_dag_info
294def _get_info_from_path(wms_path: str | os.PathLike) -> tuple[str, dict[str, dict[str, Any]], str]:
295 """Gather run information from a given run directory.
297 Parameters
298 ----------
299 wms_path : `str` or `os.PathLike`
300 Directory containing HTCondor files.
302 Returns
303 -------
304 wms_workflow_id : `str`
305 The run id which is a DAGman job id.
306 jobs : `dict` [`str`, `dict` [`str`, `~typing.Any`]]
307 Information about jobs read from files in the given directory.
308 The key is the HTCondor id and the value is a dictionary of HTCondor
309 keys and values.
310 message : `str`
311 Message to be printed with the summary report.
312 """
313 # Ensure path is absolute, in particular for folks helping
314 # debug failures that need to dig around submit files.
315 wms_path = Path(wms_path).resolve()
317 messages = []
318 try:
319 wms_workflow_id, jobs = read_dag_log(wms_path)
320 _LOG.debug("_get_info_from_path: from dag log %s = %s", wms_workflow_id, jobs)
321 _update_jobs(jobs, read_node_status(wms_path))
322 _LOG.debug("_get_info_from_path: after node status %s = %s", wms_workflow_id, jobs)
324 # Add more info for DAGman job
325 job = jobs[wms_workflow_id]
326 job.update(read_dag_status(wms_path))
328 job["total_jobs"], job["state_counts"] = _get_state_counts_from_jobs(wms_workflow_id, jobs)
329 if "bps_run" not in job:
330 _add_run_info(wms_path, job)
332 message = htc_check_dagman_output(wms_path)
333 if message:
334 messages.append(message)
335 _LOG.debug(
336 "_get_info: id = %s, total_jobs = %s", wms_workflow_id, jobs[wms_workflow_id]["total_jobs"]
337 )
339 # Add extra pieces of information which cannot be found in HTCondor
340 # generated files like 'GlobalJobId'.
341 #
342 # Do not treat absence of this file as a serious error. Neither runs
343 # submitted with earlier versions of the plugin nor the runs submitted
344 # with Pegasus plugin will have it at the moment. However, once enough
345 # time passes and Pegasus plugin will have its own report() method
346 # (instead of sneakily using HTCondor's one), the lack of that file
347 # should be treated as seriously as lack of any other file.
348 try:
349 job_info = read_dag_info(wms_path)
350 except FileNotFoundError as exc:
351 message = f"Warn: Some information may not be available: {exc}"
352 messages.append(message)
353 else:
354 schedd_name = next(iter(job_info))
355 job_ad = next(iter(job_info[schedd_name].values()))
356 job.update(job_ad)
357 except FileNotFoundError as err:
358 message = f"Could not find HTCondor files in '{wms_path}' ({err})"
359 _LOG.debug(message)
360 messages.append(message)
361 message = htc_check_dagman_output(wms_path)
362 if message:
363 messages.append(message)
364 wms_workflow_id = MISSING_ID
365 jobs = {}
367 # Add more condor_q-like info.
368 for job in jobs.values():
369 htc_tweak_log_info(wms_path, job)
371 message = "\n".join([msg for msg in messages if msg])
372 _LOG.debug("wms_workflow_id = %s, jobs = %s", wms_workflow_id, jobs.keys())
373 _LOG.debug("message = %s", message)
374 return wms_workflow_id, jobs, message
377def _create_detailed_report_from_jobs(
378 wms_workflow_id: str, jobs: dict[str, dict[str, Any]]
379) -> dict[str, WmsRunReport]:
380 """Gather run information to be used in generating summary reports.
382 Parameters
383 ----------
384 wms_workflow_id : `str`
385 The run id to create the report for.
386 jobs : `dict` [`str`, `dict` [`str`, Any]]
387 Mapping HTCondor job id to job information.
389 Returns
390 -------
391 run_reports : `dict` [`str`, `lsst.ctrl.bps.WmsRunReport`]
392 Run information for the detailed report. The key is the given HTCondor
393 id and the value is a collection of report information for that run.
394 """
395 _LOG.debug("_create_detailed_report: id = %s, job = %s", wms_workflow_id, jobs[wms_workflow_id])
397 dag_ad = jobs[wms_workflow_id]
399 report = WmsRunReport(
400 wms_id=f"{dag_ad['ClusterId']}.{dag_ad['ProcId']}",
401 global_wms_id=dag_ad.get("GlobalJobId", "MISS"),
402 path=dag_ad["Iwd"],
403 label=dag_ad.get("bps_job_label", "MISS"),
404 run=dag_ad.get("bps_run", "MISS"),
405 project=dag_ad.get("bps_project", "MISS"),
406 campaign=dag_ad.get("bps_campaign", "MISS"),
407 payload=dag_ad.get("bps_payload", "MISS"),
408 operator=_get_owner(dag_ad),
409 run_summary=_get_run_summary(dag_ad),
410 state=_htc_status_to_wms_state(dag_ad),
411 total_number_jobs=0,
412 jobs=[],
413 job_state_counts=dict.fromkeys(WmsStates, 0),
414 exit_code_summary={},
415 )
417 payload_jobs = {} # keep track for later processing
418 specific_info = WmsSpecificInfo()
419 for job_id, job_ad in jobs.items():
420 if job_ad.get("wms_node_type", WmsNodeType.UNKNOWN) in [WmsNodeType.PAYLOAD, WmsNodeType.FINAL]:
421 try:
422 name = job_ad.get("DAGNodeName", job_id)
423 wms_state = _htc_status_to_wms_state(job_ad)
424 job_report = WmsJobReport(
425 wms_id=job_id,
426 name=name,
427 label=job_ad.get("bps_job_label", pegasus_name_to_label(name)),
428 state=wms_state,
429 )
430 if job_report.label == "init":
431 job_report.label = "pipetaskInit"
432 report.job_state_counts[wms_state] += 1
433 report.jobs.append(job_report)
434 payload_jobs[job_id] = job_ad
435 except KeyError as ex:
436 _LOG.error("Job missing key '%s': %s", str(ex), job_ad)
437 raise
438 elif is_service_job(job_ad):
439 _LOG.debug(
440 "Found service job: id='%s', name='%s', label='%s', NodeStatus='%s', JobStatus='%s'",
441 job_id,
442 job_ad["DAGNodeName"],
443 job_ad.get("bps_job_label", "MISS"),
444 job_ad.get("NodeStatus", "MISS"),
445 job_ad.get("JobStatus", "MISS"),
446 )
447 _add_service_job_specific_info(job_ad, specific_info)
449 report.total_number_jobs = len(payload_jobs)
450 report.exit_code_summary = _get_exit_code_summary(payload_jobs)
451 if specific_info:
452 report.specific_info = specific_info
454 # Workflow will exit with non-zero DAG_STATUS if problem with
455 # any of the wms jobs. So change FAILED to SUCCEEDED if all
456 # payload jobs SUCCEEDED.
457 if report.total_number_jobs == report.job_state_counts[WmsStates.SUCCEEDED]:
458 report.state = WmsStates.SUCCEEDED
460 run_reports = {report.wms_id: report}
461 _LOG.debug("_create_detailed_report: run_reports = %s", run_reports)
462 return run_reports
465def _add_service_job_specific_info(job_ad: dict[str, Any], specific_info: WmsSpecificInfo) -> None:
466 """Generate report information for service job.
468 Parameters
469 ----------
470 job_ad : `dict` [`str`, `~typing.Any`]
471 Provisioning job information.
472 specific_info : `lsst.ctrl.bps.WmsSpecificInfo`
473 Where to add message.
474 """
475 status_details = ""
476 job_status = _htc_status_to_wms_state(job_ad)
478 # Service jobs in queue are deleted when DAG is done.
479 # To get accurate status, need to check other info.
480 if (
481 job_status == WmsStates.DELETED
482 and "Reason" in job_ad
483 and (
484 "Removed by DAGMan" in job_ad["Reason"]
485 or "removed because <OtherJobRemoveRequirements = DAGManJobId =?=" in job_ad["Reason"]
486 or "DAG is exiting and writing rescue file." in job_ad["Reason"]
487 )
488 ):
489 if "HoldReason" in job_ad:
490 # HoldReason exists even if released, so check.
491 if "job_released_time" in job_ad and job_ad["job_held_time"] < job_ad["job_released_time"]:
492 # If released, assume running until deleted.
493 job_status = WmsStates.SUCCEEDED
494 status_details = ""
495 else:
496 # If job held when deleted by DAGMan, still want to
497 # report hold reason
498 status_details = f"(Job was held for the following reason: {job_ad['HoldReason']})"
500 else:
501 job_status = WmsStates.SUCCEEDED
502 elif job_status == WmsStates.SUCCEEDED:
503 status_details = "(Note: Finished before workflow.)"
504 elif job_status == WmsStates.HELD:
505 status_details = f"({job_ad['HoldReason']})"
507 template = "Status of {job_name}: {status} {status_details}"
508 context = {
509 "job_name": job_ad["DAGNodeName"],
510 "status": job_status.name,
511 "status_details": status_details,
512 }
513 specific_info.add_message(template=template, context=context)
516def _summary_report(user, hist, pass_thru, schedds=None):
517 """Gather run information to be used in generating summary reports.
519 Parameters
520 ----------
521 user : `str`
522 Run lookup restricted to given user.
523 hist : `float`
524 How many previous days to search for run information.
525 pass_thru : `str`
526 Advanced users can define the HTCondor constraint to be used
527 when searching queue and history.
529 Returns
530 -------
531 run_reports : `dict` [`str`, `lsst.ctrl.bps.WmsRunReport`]
532 Run information for the summary report. The keys are HTCondor ids and
533 the values are collections of report information for each run.
534 message : `str`
535 Message to be printed with the summary report.
536 """
537 # only doing summary report so only look for dagman jobs
538 if pass_thru:
539 constraint = pass_thru
540 else:
541 # Notes:
542 # * bps_isjob == 'True' isn't getting set for DAG jobs that are
543 # manually restarted.
544 # * Any job with DAGManJobID isn't a DAG job
545 constraint = 'bps_isjob == "True" && JobUniverse == 7'
546 if user:
547 constraint += f' && (Owner == "{user}" || bps_operator == "{user}")'
549 job_info = condor_search(constraint=constraint, hist=hist, schedds=schedds)
551 # Have list of DAGMan jobs, need to get run_report info.
552 run_reports = {}
553 msg = ""
554 for jobs in job_info.values():
555 for job_id, job in jobs.items():
556 total_jobs, state_counts = _get_state_counts_from_dag_job(job)
557 # If didn't get from queue information (e.g., Kerberos bug),
558 # try reading from file.
559 if total_jobs == 0:
560 try:
561 job.update(read_dag_status(job["Iwd"]))
562 total_jobs, state_counts = _get_state_counts_from_dag_job(job)
563 except StopIteration:
564 pass # don't kill report can't find htcondor files
566 if "bps_run" not in job:
567 _add_run_info(job["Iwd"], job)
568 report = WmsRunReport(
569 wms_id=job_id,
570 global_wms_id=job["GlobalJobId"],
571 path=job["Iwd"],
572 label=job.get("bps_job_label", "MISS"),
573 run=job.get("bps_run", "MISS"),
574 project=job.get("bps_project", "MISS"),
575 campaign=job.get("bps_campaign", "MISS"),
576 payload=job.get("bps_payload", "MISS"),
577 operator=_get_owner(job),
578 run_summary=_get_run_summary(job),
579 state=_htc_status_to_wms_state(job),
580 jobs=[],
581 total_number_jobs=total_jobs,
582 job_state_counts=state_counts,
583 )
584 run_reports[report.global_wms_id] = report
586 return run_reports, msg
589def _add_run_info(wms_path, job):
590 """Find BPS run information elsewhere for runs without bps attributes.
592 Parameters
593 ----------
594 wms_path : `str`
595 Path to submit files for the run.
596 job : `dict` [`str`, `~typing.Any`]
597 HTCondor dag job information.
599 Raises
600 ------
601 StopIteration
602 If cannot find file it is looking for. Permission errors are
603 caught and job's run is marked with error.
604 """
605 path = Path(wms_path) / "jobs"
606 try:
607 subfile = next(path.glob("**/*.sub"))
608 except (StopIteration, PermissionError):
609 job["bps_run"] = "Unavailable"
610 else:
611 _LOG.debug("_add_run_info: subfile = %s", subfile)
612 try:
613 with open(subfile, encoding="utf-8") as fh:
614 for line in fh:
615 if line.startswith("+bps_"):
616 m = re.match(r"\+(bps_[^\s]+)\s*=\s*(.+)$", line)
617 if m:
618 _LOG.debug("Matching line: %s", line)
619 job[m.group(1)] = m.group(2).replace('"', "")
620 else:
621 _LOG.debug("Could not parse attribute: %s", line)
622 except PermissionError:
623 job["bps_run"] = "PermissionError"
624 _LOG.debug("After adding job = %s", job)
627def _get_owner(job):
628 """Get the owner of a dag job.
630 Parameters
631 ----------
632 job : `dict` [`str`, `~typing.Any`]
633 HTCondor dag job information.
635 Returns
636 -------
637 owner : `str`
638 Owner of the dag job.
639 """
640 owner = job.get("bps_operator", None)
641 if not owner:
642 owner = job.get("Owner", None)
643 if not owner:
644 _LOG.warning("Could not get Owner from htcondor job: %s", job)
645 owner = "MISS"
646 return owner
649def _get_run_summary(job):
650 """Get the run summary for a job.
652 Parameters
653 ----------
654 job : `dict` [`str`, `~typing.Any`]
655 HTCondor dag job information.
657 Returns
658 -------
659 summary : `str`
660 Number of jobs per PipelineTask label in approximate pipeline order.
661 Format: <label>:<count>[;<label>:<count>]+
662 """
663 summary = job.get("bps_job_summary", job.get("bps_run_summary", None))
664 if not summary:
665 summary, _, _ = summarize_dag(job["Iwd"])
666 if not summary:
667 _LOG.warning("Could not get run summary for htcondor job: %s", job)
668 _LOG.debug("_get_run_summary: summary=%s", summary)
670 # Workaround sometimes using init vs pipetaskInit
671 summary = summary.replace("init:", "pipetaskInit:")
673 if "pegasus_version" in job and "pegasus" not in summary:
674 summary += ";pegasus:0"
676 return summary
679def _get_exit_code_summary(jobs):
680 """Get the exit code summary for a run.
682 Parameters
683 ----------
684 jobs : `dict` [`str`, `dict` [`str`, Any]]
685 Mapping HTCondor job id to job information.
687 Returns
688 -------
689 summary : `dict` [`str`, `list` [`int`]]
690 Jobs' exit codes per job label.
691 """
692 summary = {}
693 for job_id, job_ad in jobs.items():
694 job_label = job_ad["bps_job_label"]
695 summary.setdefault(job_label, [])
696 try:
697 exit_code = 0
698 job_status = job_ad["JobStatus"]
699 match job_status:
700 case htcondor.JobStatus.COMPLETED | htcondor.JobStatus.HELD | htcondor.JobStatus.REMOVED:
701 exit_code = job_ad["ExitSignal"] if job_ad["ExitBySignal"] else job_ad["ExitCode"]
702 case (
703 htcondor.JobStatus.IDLE
704 | htcondor.JobStatus.RUNNING
705 | htcondor.JobStatus.TRANSFERRING_OUTPUT
706 | htcondor.JobStatus.SUSPENDED
707 ):
708 pass
709 case _:
710 _LOG.debug("Unknown 'JobStatus' value ('%d') in classad for job '%s'", job_status, job_id)
711 if exit_code != 0:
712 summary[job_label].append(exit_code)
713 except KeyError as ex:
714 _LOG.debug("Attribute '%s' not found in the classad for job '%s'", ex, job_id)
715 return summary
718def _get_state_counts_from_jobs(
719 wms_workflow_id: str, jobs: dict[str, dict[str, Any]]
720) -> tuple[int, dict[WmsStates, int]]:
721 """Count number of jobs per WMS state.
723 The workflow job and the service jobs are excluded from the count.
725 Parameters
726 ----------
727 wms_workflow_id : `str`
728 HTCondor job id.
729 jobs : `dict [`dict` [`str`, `~typing.Any`]]
730 HTCondor dag job information.
732 Returns
733 -------
734 total_count : `int`
735 Total number of dag nodes.
736 state_counts : `dict` [`lsst.ctrl.bps.WmsStates`, `int`]
737 Keys are the different WMS states and values are counts of jobs
738 that are in that WMS state.
739 """
740 state_counts = dict.fromkeys(WmsStates, 0)
741 for job_id, job_ad in jobs.items():
742 if job_id != wms_workflow_id and job_ad.get("wms_node_type", WmsNodeType.UNKNOWN) in [
743 WmsNodeType.PAYLOAD,
744 WmsNodeType.FINAL,
745 ]:
746 state_counts[_htc_status_to_wms_state(job_ad)] += 1
747 total_count = sum(state_counts.values())
749 return total_count, state_counts
752def _get_state_counts_from_dag_job(job):
753 """Count number of jobs per WMS state.
755 Parameters
756 ----------
757 job : `dict` [`str`, `~typing.Any`]
758 HTCondor dag job information.
760 Returns
761 -------
762 total_count : `int`
763 Total number of dag nodes.
764 state_counts : `dict` [`lsst.ctrl.bps.WmsStates`, `int`]
765 Keys are the different WMS states and values are counts of jobs
766 that are in that WMS state.
767 """
768 _LOG.debug("_get_state_counts_from_dag_job: job = %s %s", type(job), len(job))
769 state_counts = dict.fromkeys(WmsStates, 0)
770 if "DAG_NodesReady" in job:
771 state_counts = {
772 WmsStates.UNREADY: job.get("DAG_NodesUnready", 0),
773 WmsStates.READY: job.get("DAG_NodesReady", 0),
774 WmsStates.HELD: job.get("DAG_JobsHeld", 0),
775 WmsStates.SUCCEEDED: job.get("DAG_NodesDone", 0),
776 WmsStates.FAILED: job.get("DAG_NodesFailed", 0),
777 WmsStates.PRUNED: job.get("DAG_NodesFutile", 0),
778 WmsStates.MISFIT: job.get("DAG_NodesPre", 0) + job.get("DAG_NodesPost", 0),
779 }
780 total_jobs = job.get("DAG_NodesTotal")
781 _LOG.debug("_get_state_counts_from_dag_job: from DAG_* keys, total_jobs = %s", total_jobs)
782 elif "NodesFailed" in job:
783 state_counts = {
784 WmsStates.UNREADY: job.get("NodesUnready", 0),
785 WmsStates.READY: job.get("NodesReady", 0),
786 WmsStates.HELD: job.get("JobProcsHeld", 0),
787 WmsStates.SUCCEEDED: job.get("NodesDone", 0),
788 WmsStates.FAILED: job.get("NodesFailed", 0),
789 WmsStates.PRUNED: job.get("NodesFutile", 0),
790 WmsStates.MISFIT: job.get("NodesPre", 0) + job.get("NodesPost", 0),
791 }
792 try:
793 total_jobs = job.get("NodesTotal")
794 except KeyError as ex:
795 _LOG.error("Job missing %s. job = %s", str(ex), job)
796 raise
797 _LOG.debug("_get_state_counts_from_dag_job: from NODES* keys, total_jobs = %s", total_jobs)
798 else:
799 # With Kerberos job auth and Kerberos bug, if warning would be printed
800 # for every DAG.
801 _LOG.debug("Can't get job state counts %s", job["Iwd"])
802 total_jobs = 0
804 _LOG.debug("total_jobs = %s, state_counts: %s", total_jobs, state_counts)
805 return total_jobs, state_counts
808def _update_jobs(jobs1, jobs2):
809 """Update jobs1 with info in jobs2.
811 (Basically an update for nested dictionaries.)
813 Parameters
814 ----------
815 jobs1 : `dict` [`str`, `dict` [`str`, `~typing.Any`]]
816 HTCondor job information to be updated.
817 jobs2 : `dict` [`str`, `dict` [`str`, `~typing.Any`]]
818 Additional HTCondor job information.
819 """
820 for job_id, job_ad in jobs2.items():
821 if job_id in jobs1:
822 jobs1[job_id].update(job_ad)
823 else:
824 jobs1[job_id] = job_ad
827def is_service_job(job_ad: dict[str, Any]) -> bool:
828 """Determine if a job is a service one.
830 Parameters
831 ----------
832 job_ad : `dict` [`str`, Any]
833 Information about an HTCondor job.
835 Returns
836 -------
837 is_service_job : `bool`
838 True if the job is a service one, false otherwise.
840 Notes
841 -----
842 At the moment, HTCondor does not provide a native way to distinguish
843 between payload and service jobs in the workflow. This code depends
844 on read_node_status adding wms_node_type.
845 """
846 return job_ad.get("wms_node_type", WmsNodeType.UNKNOWN) == WmsNodeType.SERVICE