XRootD and FTS Workshop @ CNRS Rhône Auvergne Delegation Lyon, France
Amphitheatre
CNRS Delegation Rhône-Auvergne
LOGIN to see the Zoom connection details. :)
The 2026 XRootD and FTS workshop will take place on 14-18 September, 2026, at CNRS Rhône Auvergne Delegation Amphitheater located in Lyon, France. This event will be hosted by CC-IN2P3.
The XRootD and FTS workshop brings together the XRootD and FTS developers and people from Academia, Research, and Industry to discuss current and future data access activities as related to the XRootD framework and to the FTS project.
Presentations focus on achievements, shortcomings, requirements and future plans for both XRootD and FTS projects.
Registration will open on 1st April and close on 1st September. The cost for the workshop is 122€ for the five days, which includes lunches and coffee breaks. There is also the possibility to join the social dinner for which the cost has yet to be determined.
-
-
11:30 AM
→
1:00 PM
Registration Amphitheatre
Amphitheatre
CNRS Delegation Rhône-Auvergne
2 Av. Albert Einstein 69100 Villeurbanne France -
1:00 PM
→
2:00 PM
Lunch
-
2:00 PM
→
2:50 PM
Welcome & Logistics
-
2:50 PM
→
3:50 PM
FTS Presentations
-
2:50 PM
FTS in Rubin Data Transfers 30m
Rubin/LSST is using the Rucio/FTS/Xrootd/dCache ecosystem for data transfer and distribution between their Fr/UK/US data facilities and independent data access centers around the world. Here we will outline the integration of Rubin's native data management system Butler and the Rucio ecosystem. We will focus on Rubin's specific challenge of efficiently transferring large number of small files, and will discuss the lesson we learned and tuning we did during recent large scale transfers and distributions.
Speaker: Wei Yang (SLAC National Accelerator Laboratory (US)) -
3:20 PM
Operating Data Movement at SKA Scale: Progress and Future Directions for SRCNet 30m
SRCNet provides the international data movement infrastructure for the SKA Regional Centre (SRC) network, connecting geographically distributed facilities that will store, distribute, and process SKA science data. As SKAO approaches operations, validating reliable, high-throughput, and scalable data transfer services becomes a critical challenge.
This talk presents the current status of SRCNet, recent large-scale data movement tests, and lessons learned from deploying and operating a globally distributed research infrastructure. We discuss requirements driven by SKA-scale workloads, experiences with transfer technologies and data access frameworks, including FTS and XRootD-based approaches, and ongoing work to optimise performance, monitoring, and operational resilience.
We conclude with a roadmap towards production operations and highlight opportunities for collaboration between the radio astronomy and wider data-intensive science communities.
Speaker: James William Walder (Science and Technology Facilities Council STFC (GB))
-
2:50 PM
-
3:50 PM
→
4:20 PM
Coffee break
-
4:20 PM
→
5:20 PM
FTS Presentations
-
4:20 PM
Site update on FTS at RAL 20m
General overview of the current status of the FTS service being run at RAL and plans for it moving forward.
Speaker: Rose Cooper -
4:40 PM
Operational Evolution of the Fermilab FTS3 Service for non-CMS experiments: Modernizing Deployment, Automation, and Elastic Operations 20m
We present the operational evolution of the Fermilab FTS3 for non-CMS experiments, highlighting the modernization of its Kubernetes deployment, adoption of GitOps-based automation, and migration to a more maintainable infrastructure. We discuss the introduction of elastic scaling, improved monitoring, and lessons learned from operating FTS3 in production to enhance the reliability and long-term sustainability of large-scale scientific data transfers.
Speaker: Lorena Lobato Pardavila (Fermi National Accelerator Lab. (US)) -
5:00 PM
FTS Status and Future Plan at IHEP 20m
IHEP operates FTS3 as a key data transfer service for several high-energy physics experiments, including JUNO, HERD, and other related projects. This talk will report the current status of the FTS3 deployment at IHEP and summarize its use in production data transfer workflows.
We will share operational experience from deployment, maintenance, and troubleshooting, including issues observed in real transfer activities and lessons learned from daily operations. The talk will also present ongoing tests and the upgrade plan toward FTS4, covering the motivation, validation work, and expected migration considerations.
Finally, we will discuss IHEP’s preparation for future large-scale data transfer activities, especially the planned participation of JUNO and related experiments in the WLCG Data Challenge in 2027. These activities will help validate transfer workflows, assess service readiness, and prepare for future data movement requirements.
Speaker: Xuantong Zhang (Institute of High Enegry Physics, Chinese Academy of Sciences (CN))
-
4:20 PM
-
11:30 AM
→
1:00 PM
-
-
9:00 AM
→
10:15 AM
FTS Presentations
-
9:00 AM
FTS4: State of Testing Affairs 30m
Description of the FTS4 status and comprehensive EOS tests as part of DC'27 preparation.
Speaker: Mihai Patrascoiu (CERN) -
9:30 AM
FTS4@CERN: Establishing QA Procedures 45mSpeaker: Mihai Patrascoiu (CERN)
-
9:00 AM
-
10:15 AM
→
10:30 AM
Coffee break
-
10:30 AM
→
12:00 PM
FTS Presentations
-
10:30 AM
FTS4: Optimising Performance & Efficiency 30m
This talk will present the performance improvements made in FTS4 to efficiently handle thousands of concurrent transfers. It will provide an overview of how FTS manages transfers internally and how database load has been optimised to improve scalability and overall system performance.
Speaker: Louis Regnier -
11:00 AM
FTS4: Scheduler and FlowController: Scalable Scheduling and Adaptive Transfer Concurrency 30m
The FTS4 Scheduler combines fair resource allocation with database optimizations aimed at improving scalability, throughput, and reliability. The work includes optimizing database triggers and bulk updates, denormalizing frequently accessed counters, and addressing complex deadlock scenarios, including a subtle issue involving unique constraints and transaction isolation. On top of the scheduler, the new FlowController replaces the legacy FTS Optimizer with an extremum-seeking control strategy based on a feedback loop. It continuously adapts transfer concurrency based on observed endpoint behaviour, allowing FTS to improve resource utilization while reacting to overload and protecting storage endpoints.
Speaker: Nicola Pace -
11:30 AM
FTS4: Tape Daemon Re-Engineering for LHC Run 4 30m
This talk will cover the work done over the past year to re-engineer the FTS tape daemon, addressing shortcomings in the current design and improving its scalability and performance to meet the demands of LHC Run 4
Speaker: Joao Pedro Lopes (CERN)
-
10:30 AM
-
12:00 PM
→
12:15 PM
Coffee break
-
12:15 PM
→
1:15 PM
FTS Presentations
-
12:15 PM
ATLAS Community Talk 20m
ATLAS has relied on FTS to transfer millions of files across the Grid every single day. When a preproduction FTS4 instance was made available to ATLAS, we enthusiastically assisted with its testing. This talk will focus on our testing methodology, observations, and conclusions.
Speaker: Dimitrios Christidis (CERN) -
12:35 PM
LHCb <3 FTS[34] 20m
I promise I'll update the abstract with a better title/description :-)
Speaker: Christophe Haen (CERN) -
12:55 PM
CMS Community Talk 20m
An update on CMS Data Management, focused on FTS, covering updates since the last workshop and CMS's experience testing FTS4 so far.
Speaker: Christos Emmanouil (CERN)
-
12:15 PM
-
1:15 PM
→
2:30 PM
Lunch
-
2:30 PM
→
3:30 PM
FTS Presentations
-
2:30 PM
Data Challenge 2027 40m
The WLCG Data Challenge series drives change and performance improvements in our data movement systems, required in time for the start of the High-Luminosity LHC (HL-LHC). Following challenges in 2021 and 2024, this talk describes the planning in progress for the challenge in 2027, expected to take place near the start of the year. The four LHC experiments plan to take part, alongside others with data-intensive science using much of the same infrastructure and tools. Each subsequent challenge sets higher rate targets and increased complexity than the last, with the goal for DC27 sitting at 50% of the estimated HL-LHC data rates. Additional goals include scale testing of the new version of FTS (FTS4), performing all transfers using tokens, and performance read/write testing of tape systems. Plans are also in place for an XRootD load-test tool and improved XRootD monitoring to be involved in the challenge.
Speaker: Katy Ellis (Science and Technology Facilities Council STFC (GB)) -
3:10 PM
USCMS Mini-challenges: from DC24 to DC27 20m
A timeline, from the USCMS perspective, of the mini-challenges starting from the last Data Challenge in 2024 and the forecast for DC27
Speaker: Mr Diego Davila (UCSD)
-
2:30 PM
-
3:30 PM
→
4:00 PM
Coffee break
-
4:00 PM
→
5:00 PM
WLCG Open Technical Forum #13Conveners: Alessandro Di Girolamo (CERN), James Letts (Univ. of California San Diego (US))
-
4:00 PM
Introduction 10mSpeakers: Alessandro Di Girolamo (CERN), James Letts (Univ. of California San Diego (US))
-
4:10 PM
FTS status and plans 20m
FTS project presentation with status wrt what we discussed one year ago (and follow up from DC24), and plans
Speaker: Dr Maria Arsuaga Rios (CERN) -
4:30 PM
Discussion 30m
- Experiments new/different needs wrt what was discussed in the past.
- DC27: FTS goals. (scalability, stability) with the objective of understanding if FTS4 is on the right path to be capable of handling Run 4 scales.
-
4:00 PM
-
7:30 PM
→
9:30 PM
Workshop Dinner (Les Adrets - 30 Rue du Bœuf, 69005 Lyon): Workshop Group DinnerConveners: Andrew Bohdan Hanushevsky (SLAC National Accelerator Laboratory (US)), Guilherme Amadio (CERN), Dr Maria Arsuaga Rios (CERN), Mihai Patrascoiu (CERN)
-
9:00 AM
→
10:15 AM
-
-
9:00 AM
→
10:30 AM
FTS Presentations
-
9:00 AM
Rucio Project Talk 30m
The vast majority of scientific collaborations employing Rucio for their data-management needs choose FTS to orchestrate their site-to-site transfers. The two projects have collaborated closely over the years. This talk aims to cover a variety of topics, such as the migration to FTS4, the scaling for the Data Challenge 2027, and OAuth tokens. It will also mention Rucio’s current plans for migrating away from GFAL.
Speaker: Dimitrios Christidis (CERN) -
9:30 AM
Rucio/SENSE Integration update on the path to production 20m
We will present the current status of the Rucio/SENSE project and the plan to be integrated into CMS production.
Speaker: Diego Davila (UCSD) -
9:50 AM
XRootD and FTS in the MADDEN Multi-RI Data Lake 20m
The MADDEN project aims to enable seamless data access and sharing across multiple Research Infrastructures (RIs) by developing a Multi-RI Data Lake based on Rucio. As part of the project, production and test Rucio servers were deployed for the Einstein Telescope, with three Rucio Storage Elements (RSEs) located at INFN Torino, UCLouvain, and PIC. The RSEs at INFN Torino and UCLouvain are managed using XRootD. In addition, mock Cosmic Explorer Rucio infrastructures were deployed with production and test servers and two XRootD-managed RSEs at INFN Torino and UCLouvain. FTS is currently integrated into both the production and test Einstein Telescope infrastructures and is being configured for the mock Cosmic Explorer test environment. The production Einstein Telescope Rucio infrastructure also participated in the XRidge 2026 Data Challenge, providing an opportunity to evaluate the deployed services in a realistic data-transfer scenario. This contribution presents the architecture, deployment experience, and current status of XRootD and FTS within the MADDEN Multi-RI Data Lake.
Speaker: Nikita Avdeev (INFN Torino) -
10:10 AM
Deploying FTS in a Kubernetes cluster with S3-only backends 20m
The French project MesoNET, aiming at building a national federation of regional HPC/HTC centers, is deploying a Rucio proof-of-concept service to federate S3 endpoints deployed at various sites.
This deployment is using several sepcific approaches:
- The service is deployed in a Kubernetes cluster: this required to build cotnainer images for the different FTS services, reusing part of the work done for the FTS K8s tutorial
- Authentication is done with OIDC, using the MesoNET INDIGO IAM service
- All the storage backends federated by Rucio only expose a S3 endpoint, a storage technology that doesn't implement 3d party copy. The choice has been to deploy transfer nodes (based on grid storage) and configure multi-hop transfers.
The French project MesoNET, aiming at building a national federation of regional HPC/HTC centers, is deploying a Rucio proof-of-concept service to federate S3 endpoints deployed at various sites.
This deployment is using several sepcific approaches:
The service is deployed in a Kubernetes cluster: this required to build cotnainer images for the different FTS services, reusing part of the work done for the FTS K8s tutorial
Authentication is done with OIDC, using the MesoNET INDIGO IAM service
All the storage backends federated by Rucio only expose a S3 endpoint, a storage technology that doesn't implement 3d party copy. The choice has been to deploy transfer nodes (based on grid storage) and configure multi-hop transfers.Speaker: Michel Jouvin (Université Paris-Saclay (FR))
-
9:00 AM
-
10:30 AM
→
11:00 AM
Coffee break
-
11:00 AM
→
11:50 AM
FTS Presentations
-
11:00 AM
FTS4: Introducing TimescaleDB 20m
Leveraging TimescaleDB to plot FTS4 timeseries data: prototype and future direction
Speaker: Mihai Patrascoiu (CERN) -
11:20 AM
FTS Open Discussion 30m
-
11:00 AM
-
11:50 AM
→
12:00 PM
Group Photo Amphitheatre
Amphitheatre
CNRS Delegation Rhône-Auvergne
2 Av. Albert Einstein 69100 Villeurbanne France -
12:00 PM
→
12:05 PM
Welcome & Logistics
-
12:05 PM
→
1:00 PM
XRootD Presentations Amphitheatre
Amphitheatre
CNRS Delegation Rhône-Auvergne
2 Av. Albert Einstein 69100 Villeurbanne France-
12:05 PM
Catch up with XRootD! 25m
This talk rounds up new features added to XRootD since the last workshop and delves into security fixes.
Speaker: Andrew Bohdan Hanushevsky (SLAC National Accelerator Laboratory (US)) -
12:30 PM
Evolution of XRootD Monitoring: Metrics 30m
In this contribution, we will present our work to add native support for Prometheus metrics in XRootD. We will discuss the new XrdMetrics module both from the developer and user points of view, how to configure and use the new system, and its current development status.
Speaker: Guilherme Amadio (CERN)
-
12:05 PM
-
1:00 PM
→
2:30 PM
Lunch
-
2:30 PM
→
3:40 PM
XRootD Presentations Amphitheatre
Amphitheatre
CNRS Delegation Rhône-Auvergne
2 Av. Albert Einstein 69100 Villeurbanne France-
2:30 PM
Evolution of XRootD Monitoring: Logs and Traces 30m
In this contribution, we will discuss a new XRootD detailed monitoring collector (i.e. UDP packet parser/correlator) currently under development. We will discuss its features, architecture, configuration, and deployment modes to support WLCG monitoring requirements in the future.
Speaker: Guilherme Amadio (CERN) -
3:00 PM
Towards WLCG-wide XRootD monitoring: deployment plans and the road to DC27 20m
This talk gives an update on the state of a common transfer monitoring pipeline for XRootD transfers in WLCG, covering both native XRootD servers and xroot traffic over dCache. The collector is now validated and ready for large-scale deployment, and we present what deploying it at other sites involves and how we plan to drive adoption ahead of DC27.
Speaker: Panos Paparrigopoulos (CERN) -
3:20 PM
Improving the OSG XRootD Collector and Validating Against MonaLisa 20m
This talk presents improvements to the OSG XRootD monitoring collector: fixing a critical t-stream parsing bug that fabricated bogus file-close records, adding topology and SciTags resolution, new Prometheus metrics, and expanded configurability for multi-VO deployments. We conclude with a validation of the collector against MonaLisa at CERN, confirming the fixes and establishing a cross-system XRootD monitoring baseline for WLCG.
Speaker: Karanjot Singh (CERN)
-
2:30 PM
-
3:40 PM
→
4:00 PM
Coffee break
-
4:00 PM
→
5:20 PM
XRootD Presentations Amphitheatre
Amphitheatre
CNRS Delegation Rhône-Auvergne
2 Av. Albert Einstein 69100 Villeurbanne France-
4:00 PM
Global Profiling of Open Science Data Federation Services using the Pelican Platform 20m
Contemporary research relies heavily on computational resources and storage, with data sharing serving as a critical element. Data access remains a central challenge. The Open Science Data Federation (OSDF) project aims to establish a global scientific data distribution network by leveraging the Pelican Platform and the National Research Platform (NRP). OSDF is based on the XRootD and Pelican projects. Characterizing the performance boundaries of the Pelican Platform under various configurations, such as transfer rate limits, buffer settings, and geographic distribution, is essential. To this end, OSDF access was systematically profiled globally. The evaluation considered multiple file sizes, parallel data streams, and diverse client-server geographic relationships. This approach facilitates comprehensive monitoring of XRootD and Pelican performance across a range of scenarios. Results indicate that OSDF can deliver transfer speeds approaching the maximum capabilities of client systems.
Speaker: Fabio Andrijauskas (Univ. of California San Diego (US)) -
4:20 PM
Secure data access to ancient data sets: XRootD in and for data preservation 20m
Going back in time by about 30 years or more, the CERNLIB library was used by a large number of experiments and collaborations, across a huge number of platforms and architectures. Written (mostly) in C and Fortran, it contained a large number of highly optimized tools, among them the "Physics Analysis Workstation (PAW)", a popular tool to visualize scientific data at the time, and the predecessor of Root. While no longer officially supported for more than 20 years, CERNLIB is still alive and plays an important role in data preservation. This presentation tells the story about how xrootd came into the game of data preservation, and how it can be used in this context, within the original frameworks of old experiments.
Speaker: Dr Ulrich Schwickerath (CERN) -
4:40 PM
Experience of AI Assisted Code Improvement for XRootD 10m
We report here our experience of using AI to make change/improvement to open source software like XRootD. We choose three real world tasks to see if AI can have a comprehensive understanding of both XRootD software stack, other software used by Xrootd (e.g. Fuse, Scitoken), and our intention. Theses tasks are: creating a separate version of xrdadler32 that does not depend on any XRootD libraries; upgrading XrdFfs (xrootdfs) from using Fuse 2.6 to Fuse 3.x with limited functionality changes; and asking AI to help us understand how Sci-tokens are passed to the XRootD clients. In each of these tasks, AI worked incredibly well as a coding assistants - it was quick to figure out underline relations within the XRootD stack, and make code change with minimum bugs. But AI does make mistakes that require human corrections. We will point out and analyze those mistakes.
Speaker: Sarah Yang -
4:50 PM
Mirage – File-Level Simulation on an XRootD Server 30m
Mirage – File-Level Simulation on an XRootD Server
Speaker: Diogo Mattioli (CERN)
-
4:00 PM
-
9:00 AM
→
10:30 AM
-
-
8:30 AM
→
10:30 AM
XRootD Presentations Amphitheatre
Amphitheatre
CNRS Delegation Rhône-Auvergne
2 Av. Albert Einstein 69100 Villeurbanne France-
9:00 AM
XCache - the future development menu 30m
Discuss status and possible extensions of XCache, in particular related to:
- analysis facilities
- resource monitor
- cache control and cache consistency
- the Pelican projectRequired time: 30 min
Speaker: Matevz Tadel (Univ. of California San Diego (US)) -
9:30 AM
XCache setup and support in Germany 20m
Several sites supporting WLCG in Germany operate XCache services. Within the recently initiated FUSE project we prepare standardized deployment procedures which match the constraints and requirements of our sites. We are working on light-weight containers and well-tested setup examples, including documentation and a dedicated monitoring service. In addition we are preparing dedicated performance test-suites ranging from low-level FIO tests to large-scale application tests based on Hammercloud.
Speaker: Raphael Kleinemuhl (Bergische Universität Wuppertal) -
9:50 AM
Verifying replica integrity before XRootD redirects 20m
OpenVerify is an XRootD OFS plugin that probes candidate replicas before the client is redirected. It can locate via CMS Rucio with cmsd fallback, quarantine unhealthy sites, and export Prometheus metrics to reduce federated open failures on missing or broken files.
Speaker: Rahul Chauhan (University of Wisconsin (US)) -
10:10 AM
XRootD Redirectors at Fermilab: Architecture and Operations 20m
This presentation provides an operational overview of the XRootD services hosted at Fermilab, including the USCMS AAA regional and FNAL site redirectors, as well as the StashCache redirector supporting the OSG Data Federation. We summarize the current deployment architecture, service monitoring, and operational practices.
Speaker: Lorena Lobato Pardavila (Fermi National Accelerator Lab. (US))
-
9:00 AM
-
10:30 AM
→
11:00 AM
Coffee break
-
11:00 AM
→
1:00 PM
XRootD Presentations Amphitheatre
Amphitheatre
CNRS Delegation Rhône-Auvergne
2 Av. Albert Einstein 69100 Villeurbanne France-
11:00 AM
An XRootD Agent as A Unified Storage Interface for INK Platform 20m
To meet the growing and evolving needs of high-energy physics users in interactive data analysis and processing, an all-in-one platform, the Interactive Analysis Workbench (INK) is proposed and developed by IHEP CAS, as a secure and reliable bridge between users and computing clusters..
To bridge private cluster storage systems and interactive web accessing, we choose the XRootD framework combined with the FastAPI framework as a middleware to handle all filesystem access requests as well as internal access within INK. As a unified storage interface of INK, it acts as an agent to accept http access and proxy to backend server with XRootD protocol. The storage interface is one of fundamental interfaces and with it, INK has become one of the most popular interactive analysis platforms in IHEP.Speaker: Dr Yujiang BI (Institute of High Energy Physics, Chinese Academy of Sciences) -
11:20 AM
XRootD Remote Reads for WLCG Data Challenge 20m
xrd-readgen generates controlled, measured XRootD remote-read traffic against federation redirectors and site endpoints for WLCG data challenges. This talk walks through its interfaces, internal design, and client-side metrics, and invites comments and suggestions to improve the tool.
Speaker: Rahul Chauhan (University of Wisconsin (US)) -
11:40 AM
Update on OU_OSCER_ATLAS backend storage migration from XRootD to CephFS 20m
Update on OU_OSCER_ATLAS backend storage migration from XRootD to CephFS
Speaker: Horst Severini (University of Oklahoma (US)) -
12:00 PM
Real-time checksums in XRootD 20m
This talk describes the new real-time checksum capability in XRootD.
Speaker: Andrew Bohdan Hanushevsky (SLAC National Accelerator Laboratory (US)) -
12:20 PM
uCache: Transparent Client-Side Caching for XRootD 20m
High-energy physics analyses repeatedly process the same data as physicists refine algorithms and test new ideas. When the data are read from remote storage, each iteration incurs network latency and depends on shared-storage throughput, which can vary substantially under load. We present uCache (xrd-ucache), a transparent cache implemented as an XRootD client plugin that requires neither server-side deployment nor changes to analysis code. It uses local storage on the analysis machine as a cache layer between the network and memory. The cache stores only the byte ranges read by an analysis and can also rebuild cached data into a branch-aligned, recompressed form, eliminating most of the input/output and decompression costs of subsequent passes. We benchmark uCache using a realistic physics analysis of public CMS data, with both TTree and RNTuple containers, on machines ranging from a 6-core desktop to a 192-core facility node. uCache is open source under the MIT license: https://github.com/drkovalskyi/xrd-ucache
Speaker: Dmytro Kovalskyi (Massachusetts Inst. of Technology (US)) -
12:40 PM
Benchmarking and Optimisation of High-Performance Data Access and Storage Systems 20m
This CERN Openlab Summer Student project aimed to benchmark and compare different storage solutions used or considered for CERN's computing infrastructure. The main goal was to evaluate a Pure Storage appliance, HDD/SSD baselines, and CERN's production storage systems, including CephFS, XRootD and EOS. The benchmarking methodology relied on Flexible I/O (FIO) scripts designed to reasonably reproduce realistic High Energy Physics (HEP) analysis patterns and common user activities. The results of this study provide a comprehensive performance characterisation of the investigated storage solutions, helping to identify suitable candidates for the next deployments and improvements to CERN's computing services.
Speaker: Mr Gustavo Lopes (Federal University of Minas Gerais)
-
11:00 AM
-
1:00 PM
→
2:30 PM
Lunch
-
2:30 PM
→
3:30 PM
XRootD Presentations Amphitheatre
Amphitheatre
CNRS Delegation Rhône-Auvergne
2 Av. Albert Einstein 69100 Villeurbanne France-
2:30 PM
XRootD Architectures for modern storage use cases 30m
Given the rising increase in costs for procuring fast storage, this talk aims to cover some basic XRootD Architectures to handle different use cases and storage specifications, as well as a brief overview on the cost/benefit of each approach.
Speaker: Jyothish Thomas (STFC) -
3:00 PM
Pelican: Building out XRootD functionality via plugins 30m
For the Pelican Platform (https://pelicanplatform.org/), XRootD is used as the core of the data movement: whether part of the origin or cache service, when data is moved via HTTPS, XrdHttp can be found.
Doing the integration between Pelican and XRootD has required multiple new plugins into the core. They include:
- OSS storage: connecting XRootD to new storage types.
- XrdCl: providing new behavior to the caching proxy layer.
- N2N: more flexible remapings of data to alternative prefixes.
- OSS layering: adding specific functionality to the OSS stack, such as a new "persist on successful close" approach or deadlock detection.
- HTTP plugins: providing simple RESTful APIs to layers, like the cache control, that are not otherwise exported via HTTPS.
This presentation will highlight the strategy Pelican takes toward plugins and provide an overview of the new plugins we've written over the past year
Speaker: Brian Paul Bockelman (University of Wisconsin (US))
-
2:30 PM
-
3:30 PM
→
4:00 PM
Coffee break
-
4:00 PM
→
5:00 PM
WLCG Open Technical Forum #13
-
4:00 PM
introduction 10mSpeakers: Alessandro Di Girolamo (CERN), James Letts (Univ. of California San Diego (US))
-
4:10 PM
XRootD project development roadmap 20m
XRootD development roadmap: overall status report including the milestones, decisions points and risks facing the development timeline, suitable for inclusion in the WLCG Technical Roadmap, and for socialization/discussion in the session.
Speakers: Andrew Bohdan Hanushevsky (SLAC National Accelerator Laboratory (US)), Guilherme Amadio (CERN) -
4:30 PM
XRootD monitoring - status report and next steps. 30mSpeakers: Guilherme Amadio (CERN), Panos Paparrigopoulos (CERN)
-
4:00 PM
-
8:30 AM
→
10:30 AM
-
-
9:00 AM
→
10:00 AM
XRootD Presentations Amphitheatre
Amphitheatre
CNRS Delegation Rhône-Auvergne
2 Av. Albert Einstein 69100 Villeurbanne France-
9:00 AM
XrdHttp and XrdHttpTPC: The 12-Month Diff 20m
This presentation covers twelve months of evolution in the XrdHttp and XrdHttpTPC functionalities of XRootD. These include a new CORS plugin developed to address the browser-based data analysis use case of the LHCb experiment, the WLCG HTTP-TPC data integrity check initiative and how it has been implemented, and the story behind a surprising and sudden memory increase that led to Out-of-Memory kills on the production XRootD EOSCTA clusters just after the end of Run 3.
Speaker: Cedric Caffy (CERN) -
9:20 AM
HTTP/XRootD clients and ROOT - Back on the case 20m
Recent contributions to the XRootD client component have opened the possibility of using it as a multiprotocol client, supporting basic features of HTTP and S3.
This talk introduces the ongoing work to enable the ROOT framework to access HTTP and S3 servers through this XrdCl component, with the medium term goal to provide a working alternative to the Davix library.
We discuss the minimal features that are needed to achieve the objective and the ones that need to be added, trying to establish a workplan made of integration steps that cover authentication, basic features and performance-oriented ones.Speaker: Fabrizio Furano (CERN) -
9:40 AM
A Native XRootD Path for GFAL Workflows 20m
GFAL command-line tools provide a convenient, uniform interface to heterogeneous storage. For protocols already supported by XRootD, however, much of the underlying functionality exists natively but is spread across xrdfs, xrdcp, XrdCl, XrdClHttp, and the Python bindings.
This presentation describes ongoing work toward a coherent native XRootD interface for common GFAL workflows. An exploratory fsspec-based implementation served as an executable compatibility study, clarifying the required semantics, ergonomics, and testing strategy. The resulting design extends the existing XRootD stack instead of introducing another transfer engine: xrdfs provides a unified command surface, copy operations reuse the established xrdcp implementation, and reusable functionality lives in XrdCl, XrdClHttp, and the Python bindings.
The same architecture supports WLCG Tape REST operations and enables more direct Rucio and DIRAC integrations. We present the current implementation, compatibility boundaries, and remaining work.Speaker: Luis Antonio Obis Aparicio (CERN)
-
9:00 AM
-
10:00 AM
→
10:30 AM
Coffee break
-
10:30 AM
→
11:45 AM
XRootD Presentations Amphitheatre
Amphitheatre
CNRS Delegation Rhône-Auvergne
2 Av. Albert Einstein 69100 Villeurbanne France-
10:30 AM
New experimental XRootD plug-ins & R&D 30m
I will present the design of two experimental authentication plugins:
- OAUTH2 for https:// and roots:// in XRootD
- SSH authentication for roots:// in XRootD
and can optionally introduce few other R&D projects/ideas:
- XrdHttpKrb5 Kerberos Authentication support in XrdHttp
- llhttp high-performance HTTP header parsing library for XRootD/XrdHttp
- http2 support in XRootD using nghttp2
- XrdClJournalCache client plugin caching byte-ranges not blocks as a local journal, pre-configured xjcd forward proxy for http(s): and root(s):
- embeddable NFS4.1 C++ server (EOS & XRootD) with TLS, KRB5 and still missing NFS over RDMA support
- kernel-resident network filesystem client over HTTPS and RDMA usable in LAN and WAN for XRootD
Speaker: Andreas Joachim Peters (CERN) -
11:00 AM
XRootD - HTTP TPC Client support XrdClHttp 20m
XRootD - HTTP TPC Client support XrdClHttp
Speaker: Diogo Mattioli (CERN) -
11:20 AM
Terabyte Transfers over High-BDP Links: Tactics vs. Architecture 10m
Solving the large-file problem, one way or another.
Speaker: James William Walder (Science and Technology Facilities Council STFC (GB)) -
11:30 AM
XRootD Open Discussion 15m
Discuss points brought up in previous presentation and if more room left, any other topics of interest.
Speaker: Andrew Bohdan Hanushevsky (SLAC National Accelerator Laboratory (US))
-
10:30 AM
-
11:45 AM
→
12:00 PM
Workshop Wrap-up Amphitheatre
Amphitheatre
CNRS Delegation Rhône-Auvergne
2 Av. Albert Einstein 69100 Villeurbanne France -
12:00 PM
→
1:00 PM
Boxed lunch (take away) Amphitheatre
Amphitheatre
CNRS Delegation Rhône-Auvergne
2 Av. Albert Einstein 69100 Villeurbanne France
-
9:00 AM
→
10:00 AM