1.0.2

What's Changed

• Updating dependency to capture latest panda-common tag by @vippolit in #745

Full Changelog: 1.0.1...1.0.2

1.0.1

What's Changed

• Enhance task management for RSE DT and logging across multiple modules by @tmaeno in #716
• Use WN GPU monitoring as primary source for GPU brokerage attribute checks by @EdwardKaravakis in #714
• Fix == operator in GPU shorthand being parsed as === by @EdwardKaravakis in #717
• Fix microarchitecture shorthand storing ==Ampere instead of Ampere by @EdwardKaravakis in #718
• Fixed dispatchDBlock for tarBallViaDDM and disabled walltime scaling when NE is used by @tmaeno in #720
• typo by @tmaeno in #721
• Changed defaults for noChildRetry [ATLASPANDA-1747] by @fbarreir in #722
• Treat single = as == in GPU version comparisons for JSON gpu_spec by @EdwardKaravakis in #723
• Build and push latest Docker image on every merge to main by @EdwardKaravakis in #724
• Fix Docker build trigger to watch master branch by @EdwardKaravakis in #725
• Improve Postgres NOWAIT behavior in WrappedCursor: raise LockNotAvailable only when rows exist but are all locked by @tmaeno in #726
• Enable no-cache option for Docker image build and push by @tmaeno in #727
• Implement async request processing API and related functionality by @tmaeno in #728
• MCP: fixed special cases with certificates and tokens by @fbarreir in #729
• uppercase the database schema by @wguanicedew in #719
• Not to generate jobs for files that belong to a constituent dataset within an input dataset container if the dataset is available only at DT RSEs by @tmaeno in #730
• removing redundant tracebacks to reduce log pollution by @tmaeno in #731
• workflows4: Refactor workflow recheck logic and handle plugin errors by @mightqxc in #732
• per-user based access control for async process results by @tmaeno in #733
• Review of task ownership check by @fbarreir in #734
• Postprocessing by @fbarreir in #735
• to add file-progress report in get_task_details_json's return by @tmaeno in #736
• Async sleep+echo request handling and corresponding API endpoint by @tmaeno in #737
• Support model exclusion in GPU brokerage shorthand syntax by @EdwardKaravakis in #738
• fixed getTaskIDsWithCriteria_JEDI for missing statuses in JEDI_AUX_Status_MinTaskID by @tmaeno in #739
• Add SUM(nFilesMissing) to SQL query for input-file progress in TaskUt… by @tmaeno in #740
• fixed recursive lost file recovery by @tmaeno in #741
• Update Content-Encoding header to exclude log files in HTTPD config by @tmaeno in #742
• Fix subquery alias errors in duplicateFilesForReuse_JEDI on PostgreSQL by @EdwardKaravakis in #743
• Bump release version to 1.0.1 by @EdwardKaravakis in #744

Full Changelog: 1.0.0...1.0.1

1.0.0

What's Changed

• give priority to rucio account as set in env vars by @vippolit in #705
• added task_param['transferType'] by @tmaeno in #706
• Fixes after first round of MCP commissioning by @fbarreir in #707
• to respect site.maxtime even if task.maxwalltime is specified by @tmaeno in #708
• to fix empty errorDialog in some problematic tasks by @tmaeno in #709
• for panda to own secondary replicas by @tmaeno in #710
• fix(db): uppercase schema prefixes in SQL for schema translation by @EdwardKaravakis in #711
• workflows4: fix for secondary ds, add yaml wfd by @mightqxc in #712
• Fixed GenJobBroker by @tmaeno in #713
• Deletion of old API by @fbarreir in #715

New Contributors

• @vippolit made their first contribution in #705

Full Changelog: 0.8.1...1.0.0

0.8.1

What's Changed

• workflows4: PanDA native workflow engine first merge by @mightqxc in #672
• Fix GPU model brokerage: asymmetric include/exclude logic and case-insensitive matching by @EdwardKaravakis in #682
• optimize boost memory and fix site spec catchall checks by @wguanicedew in #650
• To allow low CPU efficiency jobs if it is due to high IO + to compact logging messaages in brokerages by @tmaeno in #684
• queuefiller: separate params for FC site by @mightqxc in #685
• to modify splitRule even when scout jobs were enforced to run with copy-to-scratch by @tmaeno in #686
• fixed updateJobStatus to prevent transferring -> running by @tmaeno in #687
• Refactor site limitation handling and update enumerations in JediTaskSpec for clarity and consistency by @tmaeno in #688
• Add support for extra queue parameters in SiteSpec by @tmaeno in #689
• improved prod job brokerage to take into account a resource type mismatch between task and jobs by @tmaeno in #690
• Improve ngMsg clarity by including resource type in the message format by @tmaeno in #691
• Fix hardcoded US cloud fallback in SiteMapper.getCloud() by @EdwardKaravakis in #692
• fix(docker): remove default httpd ssl.conf that conflicts with panda SSL config by @EdwardKaravakis in #695
• added max_io_intensity in job brokerage and fixed task brokerage message for destination_blacklist by @tmaeno in #696
• data carousel: update staged files immediately for to_pin requests by @mightqxc in #694
• to create pmerge jobs even when tasks throttled by @tmaeno in #698
• data carousel: update total_files during check; configurable check ti… by @mightqxc in #697
• data carousel: fix bug at taskbuffer method by @mightqxc in #699
• data carousel: add retire_unused in API by @mightqxc in #700
• data carousel: skip staging from inactive source tape by @mightqxc in #701
• fix(db): uppercase ATLAS_PANDA schema prefix in SQL to enable testbed substitution by @EdwardKaravakis in #702
• to throttle tasks based on nRunning or nQueued jobs in WatchDog by @tmaeno in #703
• panda_config.disable_file_dispatch to disable input file distribution by @tmaeno in #704

Full Changelog: 0.7.2...0.8.1

0.7.2

What's Changed

• to remove repeated input file info when sending the job to the pilot, and to check input dataset remote-availability based on dataset distribution by @tmaeno in #668
• data carousel: fix review feedback in data_carousel_module.py by @Copilot in #670
• data carousel: migrate data carousel related dbproxy methods to dedicate module by @mightqxc in #669
• queue filler: support ARM only PQs and FC PQs by @mightqxc in #671
• add check for maxNumFilesPerJob in job size calculations by @tmaeno in #673
• improve memory cleanup of job generator msg processor by @mightqxc in #674
• Fix typos in comments, docstrings, and log messages by @EdwardKaravakis in #675
• Use regexp for gpu_spec model matching, supporting both positive and exclusion patterns by @EdwardKaravakis in #677
• Set forceStaged parameter for RAW input datasets and update dataset type extraction logic by @tmaeno in #679
• Scout and zero walltime for analysis job brokerage by @fbarreir in #678
• Raise LockNotAvailable exception for row lock failures in WrappedCursor for postgres by @tmaeno in #680
• Reviewed usage of bind variables that don't start with ":" by @fbarreir in #681
• Refactor job parameters retrieval in JobGenerator to use shared config resolver by @tmaeno in #683

New Contributors

• @Copilot made their first contribution in #670

Full Changelog: 0.7.1...0.7.2

0.7.1

Dropped the WHERE comm_prodSourceLabel clause from query on PRODSYS_COMM

0.7.0

What's Changed

• added container-level rules to keep output datasets available during task execution by @tmaeno in #651
• To deal with input datasets with scopes and limited field counts in their names by @tmaeno in #653
• Add file locking in lost-file recovery to allow concurrent execution by multiple processes by @tmaeno in #654
• Architecture preference for prod job brokerage by @fbarreir in #652
• to update t_task.status and superstatus when reloading input by @tmaeno in #655
• Data Carousel: handle iDDS submission error by @mightqxc in #656
• To trigger duplication of sequence numbers even if reuseSecOnDemand is not specified by @tmaeno in #657
• To allow finished tasks getting reassigned to sites by @tmaeno in #658
• job API: get_description and get_description_incl_archive now allow GET and POST to bypass URL encoding limitations by @fbarreir in #659
• For postgres by @tmaeno in #661
• Merge production+analysis pilot streaming, assuming all queues have unified_dispatch enabled by @fbarreir in #663
• PEP8 cleanup for worker_module.py (Claude) by @fbarreir in #662
• add API of active job stats by site resource_type and prodsourcelabel; data carousel: clean up queued requests by @mightqxc in #660
• Data Carousel: cancel queued about-to-stage requests if no active related tasks, other fixes by @mightqxc in #664
• Removed wrong WHERE on comm_vo column from PRODSYS_COMMON by @fbarreir in #666

Full Changelog: 0.6.13...0.7.0

0.6.13

What's Changed

• Data Carousel: Fix to relate new tasks to done requests to prolong DDM rule lifetime by @mightqxc in #626
• fixed build job creation blocked due to inconsistent attemptNr and status of lib.tgz by @tmaeno in #627
• Consider CRIC's fairsharepolicy in analysis job brokerage by @fbarreir in #628
• fixed getTasksToExecCommand_JEDI to avoid contention by locking tasks separately by @tmaeno in #629
• to fix replica checks to see the main RSE at nucleus by @tmaeno in #630
• Data Carousel: Protection against duplicate staging rule submission by @mightqxc in #631
• Data Carousel: Fix to submit request for existing rule, set default rule lifetime, phase out Staging activity by @mightqxc in #632
• Refactor updateDatasetsToFinishTask_JEDI to improve transaction handling by @tmaeno in #633
• Data Carousel: fix for temp empty rse_set by @mightqxc in #634
• Check ncores match when using a benchmark number to protect against s… by @fbarreir in #635
• To set correct cmtConfig to cloning jobs when multiple architectures are specified in task.architecture by @tmaeno in #637
• Data Carousel: fix on request lock by @mightqxc in #636
• Data Carousel: fix incorrect no_staging when reusing DDM rule by @mightqxc in #638
• Data Carousel: submit iDDS when reusing DDM rule, distinguish staging vs staged rules, add rescue pending tasks by @mightqxc in #639
• Fix dis dataset readiness check to ensure total files are greater than zero by @tmaeno in #640
• Refactored Snakemake parser to use API for workflow creation by @tmaeno in #641
• Add file information to job_api: get_description and get_description_incl_archive by @fbarreir in #642
• Fix: Skip HTTPClientV1 URL randomization when behind real LB by @hrushikesh-kothem in #643
• Parse worker node information from job output report by @fbarreir in #644
• Fix analysis job brokerage to trigger second cycle regardless of first-cycle failures with faireshare policies etc by @tmaeno in #645
• protection against RSE-less nucleus by @tmaeno in #646
• disabled file availability check when only incomplete replicas are available at ANALY PQs by @tmaeno in #647
• for per_core_attr in PQ.catchall by @tmaeno in #648
• WLCG JWT profile support by @tmaeno in #649

New Contributors

• @hrushikesh-kothem made their first contribution in #643

Full Changelog: 0.6.12...0.6.13

0.6.12

What's Changed

• Adding PanDA MCP in Docker by @tmaeno in #625

Full Changelog: 0.6.11...0.6.12

0.6.11

What's Changed

• to share build jobs at the same site even if they are not yet finished by @tmaeno in #614
• fixed panda_server.cfg.template for json.decoder.JSONDecodeError by @tmaeno in #615
• Added PanDA MCP configuration, systemd unit file, and test client by @tmaeno in #616
• fix SSL certificate handling for OIDC in http_client.py by @tmaeno in #617
• fixed release_task_on_hold by @tmaeno in #618
• introduced auto pause in AtlasProdWatchDog by @tmaeno in #619
• DataCarousel: change with rucio list_dataset_replicas deep behavior by @mightqxc in #620
• update complete replica check to use --deep in AtlasBrokerUtils by @tmaeno in #621
• Panglia: removed old API methods by @fbarreir in #622
• Put back get_job_statistics_per_site_label_resource by @fbarreir in #623
• Pilot API: corrections in the parameter types for update_job/update_jobs_bulk by @fbarreir in #624

Full Changelog: 0.6.10...0.6.11