Recherche avancée

Sur d’autres sites (6473)

• Consent management platforms : Keys to compliance and user trust

  14 juin, par Joe

  Today’s marketing managers and data analysts face a tricky balancing act : gaining meaningful customer insights while respecting user privacy. Finding ways to navigate the maze of complex privacy regulations while managing consent at scale can be daunting. 

  Consent management platforms (CMPs) offer a solution. They allow companies to collect data ethically, manage user consent efficiently, and comply with privacy regulations like Europe’s General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA).

  This guide explains everything you need to know about CMPs : how they function, why they’re essential for data governance, and how they work hand-in-hand with analytics platforms to promote transparency and build trust with users.

  What is a consent management platform (CMP) and what is it for ?

  A consent management platform (CMP) helps organisations collect, organise, and store user consent for personal data processing purposes. In short, it’s a tool that ensures data collection respects user privacy and complies with regulations like the GDPR and CCPA.

  Without a CMP, businesses could face hefty fines and penalties for violating data privacy laws in different parts of the world. This shows how vital these tools are to all modern businesses.

  How do consent management platforms work ?

  CMPs give users a clear and straightforward way to provide explicit consent for data collection. These platforms manage both the technical aspects of consent storage and the user experience on your site or app.

  Here’s a simplified breakdown :

  • Cookie consent banners : The CMP displays a banner whenever a user visits your website. This banner explains the types of personal information collected and for what purpose.
  • User choice : The user can accept or reject cookies and trackers. They can often customise their preferences to choose which specific data types they’re willing to share.
  • Preference storage : The CMP stores the user’s choices. This information helps ensure that you only collect and process the permitted data.
  • Integration with other systems : CMPs integrate with other systems, such as analytics platforms and advertising networks, to ensure that data collection and processing comply with the user’s choices throughout the customer experience.

  

  A key feature of CMPs is their role in shaping privacy policy design. This design encompasses the layout, visual elements, and cues employed to seek user consent.

  A recent study by Karlstad University in Sweden showed that privacy policy design significantly influences user comprehension and willingness to disclose information. In other words, it affects consent rates considerably and is key to enhancing data collection.

  Importance of consent management for compliance

  As the world becomes increasingly interconnected, consent management is taking centre stage. Although it applies to all technologies and systems that gather or handle personal data, few instances are as relevant as smart homes.

  Smart home devices have unique access to our personal spaces and private lives. They represent a unique challenge to consent management since one person is potentially granting access to personal data from themselves and other people who may be inside or around the house.

  A 2023 study by the University College London and the University of Oxford pointed out that clear design principles and granular, contextual permission structures are essential in these situations.

  However, consent management isn’t just best practice. It’s a widespread legal requirement. Not meeting these requirements can result in hefty penalties and reputational damage to your organisation.

  Consent management under GDPR

  The European Union’s GDPR is a data protection law applicable to organisations that process the personal data of individuals residing in the European Economic Area (EEA). It’s based on the principle of opting in.

  The GDPR is one of the strongest data privacy laws globally. For non-compliance, fines can be up to €20 million or 4% of the company’s total global turnover (whichever is higher).

  It’s also one of the most heavily enforced privacy laws. According to enforcementtracker.com, Meta was fined €1.2 billion in 2023, with GDPR fines reaching over €2 billion that year alone. In the UK, the largest GDPR fine is €22.05 million, according to Statista. It pays to comply.

  The GDPR has specific rules around consent, including that it must be :

  • Freely given : Users must not be pressured or coerced.
  • Specific : Must be given for specific data processing purposes.
  • Informed : Users must be provided with clear and concise information.
  • Unambiguous : Permission must be granted through clear and affirmative action, such as checking a box or tapping a button.

  CMPs help you meet these requirements by providing a transparent and user-friendly way to obtain and manage consent.

  Consent management under CCPA

  The CCPA is another privacy protection law for businesses collecting personal information from California residents. It grants Californians the right to know what data is being collected about them, to prevent it from being sold, and to request its deletion.

  CMPs support CCPA compliance by enabling users to exercise their rights and ensuring transparent data collection practices.

  Managing consent under other regulatory frameworks

  In addition to the GDPR and CCPA, numerous other privacy regulations can impact your organisation. These regulations include :

  • The COPPA in the US
  • Brazil’s LGPD
  • Japan’s APPI
  • Canada’s PIPEDA.
  • Australia’s Privacy Act 1988 

  A CMP will help streamline the process by providing a clear, practical framework to ensure you meet all applicable requirements.

  Key features to look for in a CMP

  Choosing the right CMP is crucial for global business.

  Here are some key features to consider :

  Custom banners

  Consent banners are often among users’ first digital interactions with your brand. It should be clear, concise and visually appealing. Look for a CMP that allows you to :

  • Customise the banner’s design to match your website’s branding and aesthetics.
  • Control the banner’s positioning for optimal visibility.

  End-user management tools

  The CMP should also offer a user-friendly interface allowing visitors to grant, manage and withdraw consent.

  This includes customisable banners, granular permissions, and a preference centre. The latter is a dedicated space where users can manage their preferences anytime.

  Integration capabilities with existing systems

  The CMP should integrate with your existing technology stack, including your analytics platform, marketing automation tools and CRM. This integration ensures a smooth workflow and prevents data silos.

  How to select the right CMP for your organisation

  To find the perfect CMP, focus on your specific needs and priorities. Here’s a step-by-step guide to help you make an informed decision :

  Assessing organisational needs and goals

  Start by clearly defining your organisation’s requirements. Consider the following :

  • Types of data collected : What personal data do you collect (for example, cookies, IP addresses, location data) ?
  • Compliance requirements : Which privacy regulations must you comply with (GDPR, CCPA, COPPA) ?
  • Website or app complexity : How complex is your website or app in terms of user interactions and data collection points ?
  • Budget : How much are you willing to invest in a CMP ?

  Comparing features and pricing

  Once you thoroughly understand your needs, you can compare the features and pricing of various CMPs. Look for key features like :

  • Customisable banners
  • Granular options
  • Preference centre
  • Integration with existing systems
  • Analytics and reporting

  Once you’ve shortlisted a few options, compare the pricing and choose the best value for your budget. Take advantage of free trials before committing to a paid plan.

  Checking verified user reviews

  Read user reviews on platforms like G2 or Trustpilot to get an idea of the strengths and weaknesses of different CMPs. Look for reviews from similar organisations regarding size, industry and compliance requirements.

  Integration with a privacy-focused analytics platform

  A consent management platform acts as the bridge between your users and your analytics and marketing teams. It ensures user preferences are communicated to your analytics setup, so data collection and analysis align with their choices and comply with privacy regulations. 

  Finding a consent manager integration that works with your analytics setup is essential for businesses.

  Top five consent management platforms

  The CMP market is pretty competitive, with many players providing excellent solutions. According to Emergen Research, it was valued at $320.9 million in 2021 and is growing at 21.2%.

  Here are five of our top choices 

  1. usercentrics

  usercentrics is a comprehensive CMP with customisable banners, granular consent options and a preference centre.

  

  usercentrics geolocation rulesets page (Source : Usercentrics)

  This Google-certified CMP allows you to create global and regional consent rules to ensure compliance with local regulations like GDPR, CCPA and LGPD. For a smooth implementation, usercentrics provides access to a knowledgeable support team and a dedicated customer success executive.

  It’s worth noting that Usercentrics is the CMP we use here at Matomo. It helps us in our mission to collect and analyse data ethically and with a privacy-first mindset.

  • Key features : Customisable banners, granular permissions, cross-domain and cross-device capabilities, automatic website scans, reporting and analytics.
  • Pricing : Usercentrics offers a free plan and four paid subscription plans from €7 to €50 per month.

  2. Osano

  Osano is a user-friendly CMP focusing on transparency and ease of use.

  

  Osano’s main dashboard (Source : Osano)

  Osano can scan websites for tracking technologies without impacting the user experience.

  • Key features : Customisable banners, multi-language support, granular consent options, a preference centre and access to a knowledgeable team of compliance specialists.
  • Pricing : Osano offers a self-service free plan and a paid plan at $199 per month.

  3. Cookiebot

  Cookiebot is another popular CMP with numerous integration options, including Matomo and other analytics tools. 

  

  Cookiebot consent banner options (Source : Cookiebot)

  • Key features : A cookie scanner, a privacy trigger or button allowing users to change their consent settings, a consent management API and advanced analytics.
  • Pricing : Cookiebot offers a free plan and paid plans ranging from €7 to €50 per month.

  4. CookieYes

  CookieYes is well-suited for small businesses and websites with basic privacy needs. 

  

  CookieYes cookie banner options (Source : CookieYes)

  It offers various features, including multilingual support, geo-targeting, privacy policy generation, and a preference centre. CookieYes also integrates with analytics and CMS tools, making it easy to implement as part of your stack.

  • Key features : Customisable consent banners, granular consent options, preference centre, integration with Matomo, reporting and analytics.
  • Pricing : You can use CookieYes for free or subscribe to one of their three paid plans, which range from $10 to $55 per month.

  5. Tarte au Citron

  Tarte au Citron is an open-source, lightweight, and customisable CMP developed in France.

  

  (Source : Tarte au Citron)

  Its focus is on transparency and user experience. It provides many features free of charge, but many do require some technical knowledge to deploy. There’s also a paid subscription with ongoing support and faster implementation.

  Tarte au Citron integrates with Matomo, which is also open-source. If you’re building an open-source stack for your analytics, Matomo and Tarte au Citron make an excellent pair.

  • Key features : Open-source, customisable consent banners, integration with Matomo, works with over 220 services.
  • Pricing : You can deploy the open-source core for free, but Tarte au Citron offers three paid licenses starting at €190 for one year and reaching €690 for a lifetime license.

  How to implement cookie consent the right way

  Implementing cookie consent requires precision, time and effort. But doing it wrong can result in significant legal penalties and severe reputational damage, eroding user trust and impacting your brand’s standing. Here are the key dos and don’ts of consent :

  

  Provide clear and concise information

  Use plain language that is easy for anyone to understand. Avoid using technical terms or legal jargon that may confuse users.

  Prioritise transparency

  Be upfront about your data collection practices. Clearly state what data you collect, how you use it and who you share it with. Provide links to your privacy and cookie policies for users who want to learn more.

  Offer granular control

  Give users detailed control over as many of their cookie preferences as possible. Allow them to choose which categories of tracking cookies they consent to, such as strictly necessary, performance and marketing cookies.

  Implement user-friendly banners

  Ensure banners are prominently displayed, easy to understand, and use clear and concise language. Also, make sure they’re accessible to all users, including those with disabilities.

  Respect “do not track” settings

  It’s essential to honour users’ choices when they enable their “do not track” browser setting.

  Document consent

  Maintain a record of user consent. This will help you demonstrate compliance with data privacy regulations and provide evidence of user consent in case of an audit or investigation.

  Regularly review and update consent policies

  Review and update your customer consent policies regularly to ensure they comply with evolving data privacy regulations and reflect your current data collection practices.

  Cookie consent pitfalls to avoid

  Here are some common pitfalls to avoid that may lead to legal penalties, loss of user trust or inaccurate analytics :

  • Avoid lengthy and complicated explanations. Overwhelming users with dense legal jargon or overly technical details can lead to consent fatigue and reduce the likelihood of informed consent.
  • Don’t force users to accept all cookies or none. Blanket consent options violate user autonomy and fail to comply with regulations like the GDPR.
  • Don’t make information about your data collection practices hard to find. Hidden or buried privacy policies breed suspicion and erode trust.
  • Avoid pre-checking all cookie consents. Pre-checked boxes imply consent without explicit user action, which is not compliant with GDPR and similar regulations. Users must actively opt in, not out.

  Emerging consent management trends 

  Consent management is constantly evolving and driven by new technologies, regulations, and user expectations. Here are some emerging trends to watch out for in the short term :

  • Increased automation : AI and machine learning are helping automate consent management processes, making them more efficient and effective.
  • Enhanced user experience : CMPs are becoming more user-friendly, focusing on providing an intuitive experience.
  • Privacy-preserving analytics : CMPs are being integrated with privacy-preserving analytics platforms, such as Matomo, to enable organisations to gain insights into user behaviour without compromising privacy.
  • Google Consent Mode : In 2024, Google rolled out Consent Mode v2 to align with the Digital Markets Act. Due to upcoming privacy regulations, more versions may be coming soon.

  The Privacy Governance Report 2024 also highlights the increasing complexity of managing data privacy, with more than four in five privacy professionals taking on additional responsibilities in their existing roles. This trend will likely continue in the coming years as more privacy laws are enacted.

  Addressing upcoming privacy regulations

  Data privacy and user consent requirements continue to emerge and evolve. Businesses must stay informed and adapt their practices accordingly.

  

  In 2025, several new privacy regulations are going into effect, including :

  • New state-level privacy laws in eight US states :
    • Delaware (1 January 2025)
    • Iowa (1 January 2025)
    • Nebraska (1 January 2025)
    • New Hampshire (1 January 2025)
    • New Jersey (15 January 2025)
    • Tennessee (1 July 2025)
    • Minnesota (31 July 2025)
    • Maryland (1 October 2025)
  • The EU’s Artificial Intelligence Act (which will be implemented from 1 August 2024 through 2 August 2026) and other AI-focused regulations.
  • The UK Adequacy Decision Review has a deadline of 27 December 2025.

  Organisations that collect, process or otherwise handle data from Europe and the above-named US states should proactively prepare for these changes by :

  • Conducting regular privacy impact assessments
  • Reviewing consent mechanisms regularly
  • Implementing data minimisation strategies
  • Providing user-friendly privacy controls

  Future-proofing your consent management strategy

  CMPs are essential for managing consent preferences, protecting user privacy, and earning customers’ trust through transparency and ethical data practices.

  When choosing a CMP, you should consider key features such as integration capabilities, customisation options and user-friendly interfaces.

  Integrating a CMP with a privacy-first analytics solution like Matomo allows you to collect and analyse data in a way that’s compliant and respectful of user preferences. This combination helps maintain data integrity while demonstrating a strong commitment to privacy. 

  Start your 21-day free trial today.

• Clickstream Data : Definition, Use Cases, and More

  15 avril 2024, par Erin

  Gaining a deeper understanding of user behaviour — customers’ different paths, digital footprints, and engagement patterns — is crucial for providing a personalised experience and making informed marketing decisions. 

  In that sense, clickstream data, or a comprehensive record of a user’s online activities, is one of the most valuable sources of actionable insights into users’ behavioural patterns. 

  This article will cover everything marketing teams need to know about clickstream data, from the basic definition and examples to benefits, use cases, and best practices. 

  What is clickstream data ? 

  As a form of web analytics, clickstream data focuses on tracking and analysing a user’s online activity. These digital breadcrumbs offer insights into the websites the user has visited, the pages they viewed, how much time they spent on a page, and where they went next.

  

  Your clickstream pipeline can be viewed as a “roadmap” that can help you recognise consistent patterns in how users navigate your website. 

  With that said, you won’t be able to learn much by analysing clickstream data collected from one user’s session. However, a proper analysis of large clickstream datasets can provide a wealth of information about consumers’ online behaviours and trends — which marketing teams can use to make informed decisions and optimise their digital marketing strategy. 

  Clickstream data collection can serve numerous purposes, but the main goal remains the same — gaining valuable insights into visitors’ behaviours and online activities to deliver a better user experience and improve conversion likelihood. 

  Depending on the specific events you’re tracking, clickstream data can reveal the following : 

  • How visitors reach your website 
  • The terms they type into the search engine
  • The first page they land on
  • The most popular pages and sections of your website
  • The amount of time they spend on a page 
  • Which elements of the page they interact with, and in what sequence
  • The click path they take 
  • When they convert, cancel, or abandon their cart
  • Where the user goes once they leave your website

  As you can tell, once you start collecting this type of data, you’ll learn quite a bit about the user’s online journey and the different ways they engage with your website — all without including any personal details about your visitors.

  Types of clickstream data 

  While all clickstream data keeps a record of the interactions that occur while the user is navigating a website or a mobile application — or any other digital platform — it can be divided into two types : 

  • Aggregated (web traffic) data provides comprehensive insights into the total number of visits and user interactions on a digital platform — such as your website — within a given timeframe 
  • Unaggregated data is broken up into smaller segments, focusing on an individual user’s online behaviour and website interactions 

  One thing to remember is that to gain valuable insights into user behaviour and uncover sequential patterns, you need a powerful tool and access to full clickstream datasets. Matomo’s Event Tracking can provide a comprehensive view of user interactions on your website or mobile app — everything from clicking a button and completing a form to adding (or removing) products from their cart. 

  On that note, based on the specific events you’re tracking when a user visits your website, clickstream data can include : 

  • Web navigation data : referring URL, visited pages, click path, and exit page
  • User interaction data : mouse movements, click rate, scroll depth, and button clicks
  • Conversion data : form submissions, sign-ups, and transactions 
  • Temporal data : page load time, timestamps, and the date and time of day of the user’s last login 
  • Session data : duration, start, and end times and number of pages viewed per session
  • Error data : 404 errors and network or server response issues 

  Try Matomo for Free

  Get the web insights you need, without compromising data accuracy.

  Start for free

  No credit card required

  Clickstream data benefits and use cases 

  Given the actionable insights that clickstream data collection provides, it can serve a wide range of use cases — from identifying behavioural patterns and trends and examining competitors’ performance to helping marketing teams map out customer journeys and improve ROI.

  

  According to the global Clickstream Analytics Market Report 2024, some key applications of clickstream analytics include click-path optimisation, website and app optimisation, customer analysis, basket analysis, personalisation, and traffic analysis. 

  

  The behavioural patterns and user preferences revealed by clickstream analytics data can have many applications — we’ve outlined the prominent use cases below. 

  Customer journey mapping 

  Clickstream data allows you to analyse the e-commerce customer’s online journey and provides insights into how they navigate your website. With such a comprehensive view of their click path, it becomes easier to understand user behaviour at each stage — from initial awareness to conversion — identify the most effective touchpoints and fine-tune that journey to improve their conversion likelihood. 

  Identifying customer trends 

  Clickstream data analytics can also help you identify trends and behavioural patterns — the most common sequences and similarities in how users reached your website and interacted with it — especially when you can access data from many website visitors. 

  Think about it — there are many ways in which you can use these insights into the sequence of clicks and interactions and recurring patterns to your team’s advantage. 

  Here’s an example : 

  It can reveal that some pieces of content and CTAs are performing well in encouraging visitors to take action — which shows how you should optimise other pages and what you should strive to create in the future, too. 

  Preventing site abandonment 

  Cart abandonment remains a serious issue for online retailers : 

  According to a recent report, the global cart abandonment rate in the fourth quarter of 2023 was at 83%. 

  That means that roughly eight out of ten e-commerce customers will abandon their shopping carts — most commonly due to additional costs, slow website loading times and the requirement to create an account before purchasing. 

  In addition to cart abandonment predictions, clickstream data analytics can reveal the pages where most visitors tend to leave your website. These drop-off points are clear indicators that something’s not working as it should — and once you can pinpoint them, you’ll be able to address the issue and increase conversion likelihood.

  Improving marketing campaign ROI 

  As previously mentioned, clickstream data analysis provides insights into the customer journey. Still, you may not realise that you can also use this data to keep track of your marketing effectiveness. 

  Global digital ad spending continues to grow — and is expected to reach $836 billion by 2026. It’s easy to see why relying on accurate data is crucial when deciding which marketing channels to invest in. 

  You want to ensure you’re allocating your digital marketing and advertising budget to the channels — be it SEO, pay-per-click (PPC) ads, or social media campaigns — that impact driving conversions. 

  When you combine clickstream e-commerce data with conversion rates, you’ll find the latter in Matomo’s goal reports and have a solid, data-driven foundation for making better marketing decisions.

  Try Matomo for Free

  Get the web insights you need, without compromising data accuracy.

  Start for free

  No credit card required

  Delivering a better user experience (UX) 

  Clickstream data analysis allows you to identify specific “pain points” — areas of the website that are difficult to use and may cause customer frustration. 

  It’s clear how this would be beneficial to your business : 

  Once you’ve identified these pain points, you can make the necessary changes to your website’s layout and address any technical issues that users might face, improving usability and delivering a smoother experience to potential customers. 

  Collecting clickstream data : Tools and legal implications 

  Your team will need a powerful tool capable of handling clickstream analytics to reap the benefits we’ve discussed previously. But at the same time, you need to respect users’ online privacy throughout clickstream data collection.

  

  Generally speaking, there are two ways to collect data about users’ online activity — web analytics tools and server log files.

  Web analytics tools are the more commonly used solution. Specifically designed to collect and analyse website data, these tools rely on JavaScript tags that run in the browser, providing actionable insights about user behaviour. Server log files can be a gold mine of data, too — but that data is raw and unfiltered, making it much more challenging to interpret and analyse. 

  That brings us to one of the major clickstream challenges to keep in mind as you move forward — compliance.

  While Google remains a dominant player in the web analytics market, there’s one area where Matomo has a significant advantage — user privacy. 

  Matomo operates according to privacy laws — including the General Data Protection Regulation (GDPR) and California Consumer Privacy Act (CCPA), making it an ethical alternative to Google Analytics. 

  It should go without saying, but compliance with data privacy laws — the most talked-about one being the GDPR framework introduced by the EU — isn’t something you can afford to overlook. 

  The GDPR was first implemented in the EU in 2018. Since then, several fines have been issued for non-compliance — including the record fine of €1.2 billion that Meta Platforms, Inc. received in 2023 for transferring personal data of EU-based users to the US.

  Clickstream analytics data best practices 

  

  As valuable as it might be, processing large amounts of clickstream analytics data can be a complex — and, at times, overwhelming — process. 

  Here are some best practices to keep in mind when it comes to clickstream analysis : 

  Define your goals 

  It’s essential to take the time to define your goals and objectives. 

  Once you have a clear idea of what you want to learn from a given clickstream dataset and the outcomes you hope to see, it’ll be easier to narrow down your scope — rather than trying to tackle everything at once — before moving further down the clickstream pipeline. 

  Here are a few examples of goals and objectives you can set for clickstream analysis : 

  • Understanding and predicting users’ behavioural patterns 
  • Optimising marketing campaigns and ROI 
  • Attributing conversions to specific marketing touchpoints and channels

  Analyse your data 

  Collecting clickstream analytics data is only part of the equation ; what you do with raw data and how you analyse it matters. You can have the most comprehensive dataset at your disposal — but it’ll be practically worthless if you don’t have the skill set to analyse and interpret it. 

  In short, this is the stage of your clickstream pipeline where you uncover common sequences and consistent patterns in user behaviour. 

  Clickstream data analytics can extract actionable insights from large datasets using various approaches, models, and techniques. 

  Here are a few examples : 

  • If you’re working with clickstream e-commerce data, you should perform funnel or conversion analyses to track conversion rates as users move through your sales funnel. 
  • If you want to group and analyse users based on shared characteristics, you can use Matomo for cohort analysis. 
  • If your goal is to predict future trends and outcomes — conversion and cart abandonment prediction, for example — based on available data, prioritise predictive analytics.

  Try Matomo for Free

  Get the web insights you need, without compromising data accuracy.

  Start for free

  No credit card required

  Organise and visualise your data

  As you reach the end of your clickstream pipeline, you need to start thinking about how you will present and communicate your data. And what better way to do that than to transform that data into easy-to-understand visualisations ? 

  Here are a few examples of easily digestible formats that facilitate quick decision-making : 

  • User journey maps, which illustrate the exact sequence of interactions and user flow through your website 
  • Heatmaps, which serve as graphical — and typically colour-coded — representations of a website visitor’s activity 
  • Funnel analysis, which are broader at the top but get increasingly narrower towards the bottom as users flow through and drop off at different stages of the pipeline 

  Collect clickstream data with Matomo 

  Clickstream data is hard to beat when tracking the website visitor’s journey — from first to last interaction — and understanding user behaviour. By providing real-time insights, your clickstream pipeline can help you see the big picture, stay ahead of the curve and make informed decisions about your marketing efforts. 

  Matomo accurate data and compliance with GDPR and other data privacy regulations — it’s an all-in-one, ethical platform that can meet all your web analytics needs. That’s why over 1 million websites use Matomo for their web analytics.

  Try Matomo free for 21 days. No credit card required.

  Try Matomo for Free

  21 day free trial. No credit card required.

  
  

  Your email address
  

  Your website address
  

  Your Analytics subdomain will be .matomo.cloud (change)

  Choose your analytics subdomain
  

  .matomo.cloud

  I accept the terms and conditions and data processing agreement (DPA)

  
  Your information will be used to create an account on our cloud service. For more information please consult our privacy policy.
  
  
  

  Start improving your websites now

  

• ffmpeg : fps drop when one -map udp output unreachable [closed]

  14 mai 2024, par user25041039

  I stream a video from a raspberry pi (server) to a splitscreen of 5x5 devices (clients) through an ethernet LAN.

  


  Server

  


  On the server side, I use the following ffmpeg command that :

  


  

  1. reads a video list in loop ;

  


  2. splits it into 25 different streams ;

  


  3. maps each stream to a device through udp.

  


  


  I also have several options to minimize latency and keep the different subscreens in sync.

  


  ffmpeg -loglevel repeat+level+verbose -re -copyts -start_at_zero -rtbufsize 100000k \
    -stream_loop -1 -f concat -i stream.lst -an \
    -filter_complex "\
    [0]crop=iw/5:ih/5:0*iw/5:0*ih/5[11];
    [0]crop=iw/5:ih/5:1*iw/5:0*ih/5[12];
    [...] # truncated for readability
    [0]crop=iw/5:ih/5:4*iw/5:4*ih/5[55]" \
    -map '[11]?' -flush_packets 1 -preset ultrafast -vcodec libx264 -tune zerolatency -f mpegts "udp://100.64.0.11:1234" \
    -map '[12]?' -flush_packets 1 -preset ultrafast -vcodec libx264 -tune zerolatency -f mpegts "udp://100.64.0.12:1234" \
    [...] # truncated for readability
    -map '[55]?' -flush_packets 1 -preset ultrafast -vcodec libx264 -tune zerolatency -f mpegts "udp://100.64.0.55:1234"


  


  Clients

  


  On the client side, I use mpv to read the stream and display it. There are also options for low-latency.

  


  mpv --no-cache --force-seekable=yes --profile=low-latency --untimed --no-audio --video-rotate=90 --fs --no-config --vo=gpu --hwdec=auto udp://100.64.0.1:1234/


  


  My problem

  


  When a device is unreachable through the LAN (eg : powered down), the FPS stated by ffmpeg drops after a short period ( 10 seconds), and the stream is laggy (= some frames, then pause for 1s, ...). What I expect is the stream to go on normally, just having one of the subscreens black.

  


  Here is the full log of ffmpeg when I start the stream normally then power down one of the clients after 15s.

  


  [info] ffmpeg version 5.1.4-0+rpt3+deb12u1 Copyright (c) 2000-2023 the FFmpeg developers
[info]   built with gcc 12 (Debian 12.2.0-14)
[info]   configuration: --prefix=/usr --extra-version=0+rpt3+deb12u1 --toolchain=hardened --incdir=/usr/include/aarch64-linux-gnu --enable-gpl --disable-stripping --disable-mmal --enable-gnutls --enable-ladspa --enable-libaom --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libcodec2 --enable-libdav1d --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libglslang --enable-libgme --enable-libgsm --enable-libjack --enable-libmp3lame --enable-libmysofa --enable-libopenjpeg --enable-libopenmpt --enable-libopus --enable-libpulse --enable-librabbitmq --enable-librist --enable-librubberband --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libssh --enable-libsvtav1 --enable-libtheora --enable-libtwolame --enable-libvidstab --enable-libvorbis --enable-libvpx --enable-libwebp --enable-libx265 --enable-libxml2 --enable-libxvid --enable-libzimg --enable-libzmq --enable-libzvbi --enable-lv2 --enable-omx --enable-openal --enable-opencl --enable-opengl --enable-sand --enable-sdl2 --disable-sndio --enable-libjxl --enable-neon --enable-v4l2-request --enable-libudev --enable-epoxy --libdir=/usr/lib/aarch64-linux-gnu --arch=arm64 --enable-pocketsphinx --enable-librsvg --enable-libdc1394 --enable-libdrm --enable-vout-drm --enable-libiec61883 --enable-chromaprint --enable-frei0r --enable-libx264 --enable-libplacebo --enable-librav1e --enable-shared
[info]   libavutil      57. 28.100 / 57. 28.100
[info]   libavcodec     59. 37.100 / 59. 37.100
[info]   libavformat    59. 27.100 / 59. 27.100
[info]   libavdevice    59.  7.100 / 59.  7.100
[info]   libavfilter     8. 44.100 /  8. 44.100
[info]   libswscale      6.  7.100 /  6.  7.100
[info]   libswresample   4.  7.100 /  4.  7.100
[info]   libpostproc    56.  6.100 / 56.  6.100
[h264 @ 0x5555f1407de0] [verbose] Reinit context to 1920x1088, pix_fmt: yuv420p
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x5555f1407720] [info] Auto-inserting h264_mp4toannexb bitstream filter
[h264 @ 0x5555f140eca0] [verbose] Reinit context to 1920x1088, pix_fmt: yuv420p
[info] Input #0, concat, from 'stream.lst':
[info]   Duration: N/A, start: 0.000000, bitrate: 47 kb/s
[info]   Stream #0:0(und): Video: h264 (Main), 1 reference frame (avc1 / 0x31637661), yuv420p(tv, bt709, progressive, left), 1920x1080 (1920x1088), 47 kb/s, 24 fps, 24 tbr, 12288 tbn
[info]     Metadata:
[info]       handler_name    : Core Media Video
[info]       vendor_id       : [0][0][0][0]
[info] Stream mapping:
[info]   Stream #0:0 (h264) -> crop:default
... truncated
[info]   Stream #0:0 (h264) -> crop:default
[info]   crop:default -> Stream #0:0 (libx264)
... truncated
[info]   crop:default -> Stream #24:0 (libx264)
[info] Press [q] to stop, [?] for help
[h264 @ 0x5555f13fe050] [verbose] Reinit context to 1920x1088, pix_fmt: yuv420p
[graph 0 input from stream 0:0 @ 0x5555f1dfad40] [verbose] w:1920 h:1080 pixfmt:yuv420p tb:1/12288 fr:24/1 sar:0/1
... truncated
[graph 0 input from stream 0:0 @ 0x5555f1e01a50] [verbose] w:1920 h:1080 pixfmt:yuv420p tb:1/12288 fr:24/1 sar:0/1
[Parsed_crop_24 @ 0x5555f1dfa860] [verbose] w:1920 h:1080 sar:0/1 -> w:384 h:216 sar:0/1
... truncated
[Parsed_crop_0 @ 0x5555f1df23e0] [verbose] w:1920 h:1080 sar:0/1 -> w:384 h:216 sar:0/1
[libx264 @ 0x5555f147c220] [info] using cpu capabilities: ARMv8 NEON
[libx264 @ 0x5555f147c220] [info] profile Constrained Baseline, level 1.3, 4:2:0, 8-bit
[mpegts @ 0x5555f148ab70] [verbose] service 1 using PCR in pid=256, pcr_period=83ms
[mpegts @ 0x5555f148ab70] [verbose] muxrate VBR, sdt every 500 ms, pat/pmt every 100 ms
[info] Output #0, mpegts, to 'udp://100.64.0.11:1234':
[info]   Metadata:
[info]     encoder         : Lavf59.27.100
[info]   Stream #0:0: Video: h264, 1 reference frame, yuv420p(tv, bt709, progressive, left), 384x216 (0x0), q=2-31, 24 fps, 90k tbn
[info]     Metadata:
[info]       encoder         : Lavc59.37.100 libx264
[info]     Side data:
[info]       cpb: bitrate max/min/avg: 0/0/0 buffer size: 0 vbv_delay: N/A
[libx264 @ 0x5555f1428760] [info] using cpu capabilities: ARMv8 NEON
[libx264 @ 0x5555f1428760] [info] profile Constrained Baseline, level 1.3, 4:2:0, 8-bit
[mpegts @ 0x5555f148b080] [verbose] service 1 using PCR in pid=256, pcr_period=83ms
[mpegts @ 0x5555f148b080] [verbose] muxrate VBR, sdt every 500 ms, pat/pmt every 100 ms

... truncated
[info] Output #24, mpegts, to 'udp://100.64.0.55:1234':
[info]   Metadata:
[info]     encoder         : Lavf59.27.100
[info]   Stream #24:0: Video: h264, 1 reference frame, yuv420p(tv, bt709, progressive, left), 384x216 (0x0), q=2-31, 24 fps, 90k tbn
[info]     Metadata:
[info]       encoder         : Lavc59.37.100 libx264
[info]     Side data:
[info]       cpb: bitrate max/min/avg: 0/0/0 buffer size: 0 vbv_delay: N/A
[info] frame=    1 fps=0.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 size=   [info] frame=    9 fps=0.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=13.0 q=12.0 size=   [info] frame=   22 fps= 21 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=   34 fps= 22 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=   46 fps= 22 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=   58 fps= 23 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [AVIOContext @ 0x5555f140fa70] [verbose] Statistics: 19517 bytes read, 0 seeks
[h264 @ 0x5555f1407de0] [verbose] Reinit context to 1920x1088, pix_fmt: yuv420p
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x5555f1407720] [info] Auto-inserting h264_mp4toannexb bitstream filter
[info] frame=   70 fps= 23 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=   83 fps= 23 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=   95 fps= 23 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  107 fps= 23 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  119 fps= 23 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  132 fps= 23 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [AVIOContext @ 0x5555f140fa70] [verbose] Statistics: 19240 bytes read, 0 seeks
[h264 @ 0x5555f1407de0] [verbose] Reinit context to 1920x1088, pix_fmt: yuv420p
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x5555f1407720] [info] Auto-inserting h264_mp4toannexb bitstream filter
[info] frame=  144 fps= 23 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  156 fps= 23 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  168 fps= 23 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  180 fps= 23 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  193 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=13.0 q=12.0 q=12.0 q=27.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  205 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=13.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [AVIOContext @ 0x5555f140fa70] [verbose] Statistics: 19218 bytes read, 0 seeks
[h264 @ 0x5555f1407de0] [verbose] Reinit context to 1920x1088, pix_fmt: yuv420p
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x5555f1407720] [info] Auto-inserting h264_mp4toannexb bitstream filter
[h264 @ 0x5555f17af150] [verbose] Reinit context to 1920x1088, pix_fmt: yuv420p
[concat @ 0x5555f13febf0] [warning] New audio stream 0:1 at pos:68549 and DTS:8.99977s
[info] frame=  217 fps= 24 q=14.0 q=20.0 q=14.0 q=26.0 q=25.0 q=17.0 q=19.0 q=13.0 q=25.0 q=25.0 q=22.0 q=19.0 q=14.0 q=25.0 q=22.0 q=22.0 q=19.0 q=14.0 q=22.0 q=13.0 q=23.0 q=20.0 q=13.0 q=22.0 q=14.0 size=   [info] frame=  229 fps= 24 q=14.0 q=22.0 q=14.0 q=26.0 q=24.0 q=18.0 q=21.0 q=15.0 q=25.0 q=24.0 q=20.0 q=22.0 q=15.0 q=25.0 q=21.0 q=21.0 q=22.0 q=15.0 q=23.0 q=12.0 q=22.0 q=22.0 q=12.0 q=23.0 q=14.0 size=   [info] frame=  241 fps= 24 q=16.0 q=22.0 q=15.0 q=26.0 q=25.0 q=17.0 q=22.0 q=18.0 q=25.0 q=24.0 q=20.0 q=22.0 q=17.0 q=25.0 q=22.0 q=21.0 q=23.0 q=17.0 q=23.0 q=15.0 q=21.0 q=23.0 q=17.0 q=23.0 q=15.0 size=   [info] frame=  253 fps= 24 q=18.0 q=24.0 q=19.0 q=26.0 q=25.0 q=17.0 q=23.0 q=20.0 q=26.0 q=25.0 q=21.0 q=23.0 q=19.0 q=26.0 q=23.0 q=22.0 q=22.0 q=19.0 q=23.0 q=15.0 q=23.0 q=22.0 q=19.0 q=22.0 q=16.0 size=   [info] frame=  265 fps= 24 q=20.0 q=23.0 q=19.0 q=27.0 q=26.0 q=19.0 q=25.0 q=18.0 q=27.0 q=25.0 q=23.0 q=25.0 q=18.0 q=26.0 q=22.0 q=24.0 q=23.0 q=17.0 q=22.0 q=14.0 q=16.0 q=19.0 q=14.0 q=18.0 q=14.0 size=   [info] frame=  278 fps= 24 q=12.0 q=22.0 q=25.0 q=21.0 q=15.0 q=12.0 q=24.0 q=24.0 q=25.0 q=13.0 q=12.0 q=25.0 q=23.0 q=24.0 q=12.0 q=12.0 q=25.0 q=26.0 q=24.0 q=12.0 q=12.0 q=17.0 q=22.0 q=17.0 q=12.0 size=   [info] frame=  290 fps= 24 q=13.0 q=22.0 q=22.0 q=22.0 q=16.0 q=18.0 q=22.0 q=22.0 q=22.0 q=18.0 q=19.0 q=22.0 q=22.0 q=21.0 q=18.0 q=16.0 q=22.0 q=21.0 q=21.0 q=17.0 q=12.0 q=22.0 q=21.0 q=22.0 q=14.0 size=   [info] frame=  302 fps= 24 q=18.0 q=21.0 q=22.0 q=21.0 q=19.0 q=20.0 q=21.0 q=21.0 q=21.0 q=20.0 q=20.0 q=21.0 q=21.0 q=21.0 q=19.0 q=20.0 q=21.0 q=21.0 q=22.0 q=19.0 q=17.0 q=22.0 q=21.0 q=21.0 q=17.0 size=   [info] frame=  314 fps= 24 q=19.0 q=21.0 q=21.0 q=21.0 q=20.0 q=20.0 q=21.0 q=21.0 q=21.0 q=20.0 q=20.0 q=21.0 q=21.0 q=21.0 q=20.0 q=20.0 q=21.0 q=20.0 q=21.0 q=20.0 q=19.0 q=21.0 q=21.0 q=21.0 q=19.0 size=   [info] frame=  326 fps= 24 q=20.0 q=21.0 q=21.0 q=20.0 q=21.0 q=21.0 q=21.0 q=20.0 q=21.0 q=21.0 q=21.0 q=21.0 q=20.0 q=21.0 q=21.0 q=21.0 q=21.0 q=20.0 q=21.0 q=21.0 q=20.0 q=21.0 q=21.0 q=21.0 q=20.0 size=   [info] frame=  339 fps= 24 q=20.0 q=20.0 q=21.0 q=20.0 q=21.0 q=21.0 q=20.0 q=19.0 q=21.0 q=21.0 q=21.0 q=21.0 q=20.0 q=20.0 q=21.0 q=20.0 q=21.0 q=20.0 q=20.0 q=21.0 q=20.0 q=21.0 q=20.0 q=21.0 q=20.0 size=   [info] frame=  351 fps= 24 q=21.0 q=20.0 q=21.0 q=20.0 q=20.0 q=20.0 q=20.0 q=19.0 q=21.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=21.0 q=21.0 q=21.0 q=20.0 q=20.0 q=20.0 q=21.0 q=21.0 q=21.0 q=21.0 q=20.0 size=   [info] frame=  363 fps= 24 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=19.0 q=20.0 q=20.0 q=21.0 q=20.0 q=20.0 q=20.0 q=21.0 q=21.0 q=21.0 q=20.0 q=20.0 q=21.0 q=21.0 q=21.0 q=20.0 q=21.0 q=21.0 size=   [info] frame=  375 fps= 24 q=20.0 q=20.0 q=20.0 q=19.0 q=20.0 q=20.0 q=20.0 q=19.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=19.0 q=21.0 q=21.0 q=20.0 q=20.0 q=20.0 q=21.0 q=20.0 q=21.0 q=20.0 q=21.0 q=21.0 size=   [info] frame=  387 fps= 24 q=20.0 q=20.0 q=20.0 q=19.0 q=20.0 q=20.0 q=19.0 q=19.0 q=20.0 q=20.0 q=20.0 q=20.0 q=20.0 q=19.0 q=21.0 q=21.0 q=20.0 q=19.0 q=19.0 q=20.0 q=21.0 q=21.0 q=19.0 q=21.0 q=20.0 size=   [info] frame=  400 fps= 24 q=20.0 q=20.0 q=20.0 q=19.0 q=20.0 q=20.0 q=18.0 q=18.0 q=19.0 q=20.0 q=20.0 q=19.0 q=20.0 q=19.0 q=20.0 q=20.0 q=20.0 q=19.0 q=19.0 q=20.0 q=20.0 q=21.0 q=19.0 q=19.0 q=20.0 size=   [info] frame=  412 fps= 24 q=20.0 q=19.0 q=19.0 q=19.0 q=19.0 q=20.0 q=18.0 q=18.0 q=19.0 q=19.0 q=20.0 q=19.0 q=20.0 q=19.0 q=19.0 q=20.0 q=20.0 q=19.0 q=19.0 q=19.0 q=20.0 q=21.0 q=19.0 q=19.0 q=19.0 size=   [info] frame=  424 fps= 24 q=20.0 q=19.0 q=20.0 q=20.0 q=19.0 q=20.0 q=18.0 q=19.0 q=19.0 q=19.0 q=20.0 q=19.0 q=20.0 q=19.0 q=19.0 q=20.0 q=20.0 q=19.0 q=19.0 q=19.0 q=21.0 q=21.0 q=19.0 q=19.0 q=19.0 size=   [info] frame=  436 fps= 24 q=20.0 q=18.0 q=19.0 q=20.0 q=18.0 q=20.0 q=18.0 q=18.0 q=19.0 q=18.0 q=20.0 q=19.0 q=20.0 q=19.0 q=19.0 q=20.0 q=19.0 q=19.0 q=19.0 q=18.0 q=21.0 q=21.0 q=19.0 q=19.0 q=19.0 size=   [info] frame=  448 fps= 24 q=19.0 q=18.0 q=19.0 q=20.0 q=18.0 q=19.0 q=18.0 q=18.0 q=19.0 q=18.0 q=19.0 q=19.0 q=20.0 q=19.0 q=18.0 q=20.0 q=19.0 q=19.0 q=19.0 q=18.0 q=21.0 q=21.0 q=19.0 q=19.0 q=19.0 size=   [info] frame=  460 fps= 24 q=19.0 q=18.0 q=19.0 q=20.0 q=18.0 q=19.0 q=18.0 q=18.0 q=19.0 q=18.0 q=19.0 q=19.0 q=20.0 q=19.0 q=18.0 q=20.0 q=19.0 q=19.0 q=20.0 q=18.0 q=20.0 q=21.0 q=19.0 q=19.0 q=18.0 size=   [info] frame=  472 fps= 24 q=19.0 q=18.0 q=19.0 q=20.0 q=18.0 q=19.0 q=18.0 q=18.0 q=19.0 q=18.0 q=19.0 q=19.0 q=19.0 q=19.0 q=18.0 q=19.0 q=19.0 q=19.0 q=20.0 q=18.0 q=20.0 q=21.0 q=19.0 q=19.0 q=18.0 size=   [info] frame=  484 fps= 24 q=19.0 q=18.0 q=19.0 q=20.0 q=18.0 q=19.0 q=18.0 q=18.0 q=19.0 q=18.0 q=19.0 q=19.0 q=19.0 q=19.0 q=18.0 q=19.0 q=19.0 q=19.0 q=19.0 q=18.0 q=20.0 q=21.0 q=19.0 q=20.0 q=18.0 size=   [info] frame=  496 fps= 24 q=21.0 q=20.0 q=21.0 q=21.0 q=18.0 q=20.0 q=21.0 q=21.0 q=21.0 q=18.0 q=20.0 q=21.0 q=22.0 q=21.0 q=18.0 q=21.0 q=21.0 q=21.0 q=21.0 q=18.0 q=21.0 q=21.0 q=21.0 q=20.0 q=19.0 size=   [info] frame=  509 fps= 24 q=21.0 q=20.0 q=22.0 q=21.0 q=17.0 q=21.0 q=22.0 q=22.0 q=22.0 q=18.0 q=20.0 q=22.0 q=22.0 q=22.0 q=17.0 q=21.0 q=22.0 q=22.0 q=22.0 q=18.0 q=21.0 q=22.0 q=22.0 q=20.0 q=18.0 size=   [info] frame=  521 fps= 24 q=20.0 q=21.0 q=22.0 q=21.0 q=17.0 q=21.0 q=22.0 q=23.0 q=22.0 q=17.0 q=20.0 q=22.0 q=23.0 q=22.0 q=17.0 q=21.0 q=22.0 q=22.0 q=22.0 q=17.0 q=21.0 q=22.0 q=22.0 q=21.0 q=18.0 size=   [info] frame=  533 fps= 24 q=20.0 q=20.0 q=22.0 q=21.0 q=15.0 q=20.0 q=22.0 q=23.0 q=22.0 q=16.0 q=19.0 q=22.0 q=23.0 q=22.0 q=16.0 q=21.0 q=22.0 q=22.0 q=22.0 q=16.0 q=21.0 q=22.0 q=22.0 q=21.0 q=17.0 size=   [AVIOContext @ 0x5555f140fa70] [verbose] Statistics: 3205712 bytes read, 2 seeks
[h264 @ 0x5555f1df2c00] [verbose] Reinit context to 1920x1088, pix_fmt: yuv420p
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x5555f1407720] [info] Auto-inserting h264_mp4toannexb bitstream filter
[h264 @ 0x5555f1949030] [verbose] Reinit context to 1920x1088, pix_fmt: yuv420p
[info] frame=  546 fps= 24 q=16.0 q=17.0 q=18.0 q=18.0 q=13.0 q=17.0 q=18.0 q=18.0 q=18.0 q=13.0 q=16.0 q=18.0 q=19.0 q=18.0 q=14.0 q=17.0 q=19.0 q=18.0 q=18.0 q=14.0 q=18.0 q=19.0 q=18.0 q=18.0 q=14.0 size=   [info] frame=  558 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=14.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  570 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  582 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  594 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  606 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  618 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  631 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  643 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  655 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [AVIOContext @ 0x5555f140fa70] [verbose] Statistics: 120365 bytes read, 2 seeks
[h264 @ 0x5555f1df2c00] [verbose] Reinit context to 1920x1088, pix_fmt: yuv420p
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x5555f1407720] [info] Auto-inserting h264_mp4toannexb bitstream filter
[h264 @ 0x5555f1949030] [verbose] Reinit context to 1920x1088, pix_fmt: yuv420p
[info] frame=  667 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  679 fps= 24 q=20.0 q=12.0 q=12.0 q=12.0 q=12.0 q=20.0 q=12.0 q=25.0 q=12.0 q=12.0 q=21.0 q=20.0 q=24.0 q=12.0 q=12.0 q=20.0 q=19.0 q=22.0 q=13.0 q=13.0 q=13.0 q=14.0 q=17.0 q=12.0 q=12.0 size=   [info] frame=  691 fps= 24 q=17.0 q=12.0 q=21.0 q=12.0 q=12.0 q=15.0 q=13.0 q=23.0 q=12.0 q=12.0 q=16.0 q=19.0 q=21.0 q=17.0 q=12.0 q=15.0 q=12.0 q=21.0 q=13.0 q=13.0 q=12.0 q=12.0 q=13.0 q=12.0 q=12.0 size=   [info] frame=  703 fps= 24 q=19.0 q=12.0 q=21.0 q=12.0 q=12.0 q=16.0 q=12.0 q=23.0 q=12.0 q=12.0 q=14.0 q=17.0 q=21.0 q=15.0 q=12.0 q=13.0 q=12.0 q=20.0 q=13.0 q=13.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  715 fps= 24 q=20.0 q=12.0 q=20.0 q=15.0 q=12.0 q=18.0 q=22.0 q=23.0 q=23.0 q=12.0 q=16.0 q=17.0 q=22.0 q=17.0 q=12.0 q=15.0 q=21.0 q=22.0 q=13.0 q=13.0 q=12.0 q=14.0 q=13.0 q=12.0 q=12.0 size=   [info] frame=  728 fps= 24 q=18.0 q=12.0 q=12.0 q=12.0 q=12.0 q=16.0 q=14.0 q=23.0 q=12.0 q=12.0 q=16.0 q=19.0 q=22.0 q=13.0 q=12.0 q=15.0 q=19.0 q=21.0 q=13.0 q=12.0 q=12.0 q=12.0 q=14.0 q=12.0 q=12.0 size=   [info] frame=  740 fps= 24 q=17.0 q=12.0 q=12.0 q=12.0 q=12.0 q=15.0 q=14.0 q=22.0 q=12.0 q=12.0 q=14.0 q=18.0 q=21.0 q=12.0 q=12.0 q=15.0 q=19.0 q=20.0 q=13.0 q=13.0 q=12.0 q=13.0 q=16.0 q=12.0 q=12.0 size=   [info] frame=  752 fps= 24 q=23.0 q=14.0 q=14.0 q=14.0 q=14.0 q=21.0 q=16.0 q=25.0 q=14.0 q=14.0 q=21.0 q=20.0 q=23.0 q=16.0 q=20.0 q=22.0 q=21.0 q=23.0 q=22.0 q=20.0 q=22.0 q=24.0 q=24.0 q=23.0 q=23.0 size=   [info] frame=  764 fps= 24 q=16.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=14.0 q=23.0 q=12.0 q=12.0 q=14.0 q=18.0 q=22.0 q=12.0 q=12.0 q=13.0 q=16.0 q=18.0 q=13.0 q=13.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  776 fps= 24 q=21.0 q=14.0 q=19.0 q=12.0 q=12.0 q=20.0 q=23.0 q=23.0 q=19.0 q=12.0 q=15.0 q=23.0 q=23.0 q=21.0 q=12.0 q=16.0 q=16.0 q=20.0 q=13.0 q=13.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  788 fps= 24 q=22.0 q=12.0 q=19.0 q=18.0 q=13.0 q=21.0 q=12.0 q=23.0 q=24.0 q=16.0 q=19.0 q=12.0 q=24.0 q=26.0 q=23.0 q=15.0 q=12.0 q=27.0 q=25.0 q=16.0 q=13.0 q=13.0 q=16.0 q=12.0 q=12.0 size=   [info] frame=  800 fps= 24 q=21.0 q=12.0 q=12.0 q=12.0 q=12.0 q=18.0 q=12.0 q=12.0 q=12.0 q=12.0 q=13.0 q=12.0 q=15.0 q=15.0 q=12.0 q=13.0 q=12.0 q=14.0 q=17.0 q=13.0 q=13.0 q=12.0 q=13.0 q=12.0 q=12.0 size=   [AVIOContext @ 0x5555f140fa70] [verbose] Statistics: 984786 bytes read, 2 seeks
[h264 @ 0x5555f1407de0] [verbose] Reinit context to 1920x1088, pix_fmt: yuv420p
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x5555f1407720] [info] Auto-inserting h264_mp4toannexb bitstream filter
[h264 @ 0x5555f187c0c0] [verbose] Reinit context to 1920x1088, pix_fmt: yuv420p
[info] frame=  812 fps= 24 q=21.0 q=12.0 q=12.0 q=12.0 q=12.0 q=18.0 q=13.0 q=12.0 q=12.0 q=12.0 q=13.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=16.0 q=15.0 q=12.0 q=12.0 q=12.0 q=13.0 q=16.0 q=12.0 size=   [info] frame=  824 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  836 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  848 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  861 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=13.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=24.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  873 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  885 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=23.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=13.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  897 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  909 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=13.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=25.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  922 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  934 fps= 24 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  938 fps= 23 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [AVIOContext @ 0x5555f140fa70] [verbose] Statistics: 29154 bytes read, 0 seeks
[h264 @ 0x5555f1407de0] [verbose] Reinit context to 1920x1088, pix_fmt: yuv420p
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x5555f1407720] [info] Auto-inserting h264_mp4toannexb bitstream filter
[h264 @ 0x5555f17af150] [verbose] Reinit context to 1920x1088, pix_fmt: yuv420p
[info] frame=  963 fps= 23 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=15.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  967 fps= 22 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=17.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  994 fps= 22 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=15.0 q=19.0 q=12.0 q=12.0 q=12.0 q=17.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame=  997 fps= 21 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=15.0 q=19.0 q=12.0 q=12.0 q=12.0 q=18.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame= 1023 fps= 21 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=14.0 q=20.0 q=18.0 q=15.0 q=15.0 q=14.0 q=13.0 q=13.0 q=13.0 q=13.0 q=13.0 q=13.0 size=   [info] frame= 1029 fps= 20 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=14.0 q=20.0 q=15.0 q=14.0 q=15.0 q=15.0 q=13.0 q=13.0 q=14.0 q=14.0 q=14.0 q=13.0 size=   [info] frame= 1055 fps= 21 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=15.0 q=19.0 q=15.0 q=14.0 q=15.0 q=16.0 q=12.0 q=13.0 q=13.0 q=13.0 q=13.0 q=12.0 size=   [info] frame= 1060 fps= 20 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=16.0 q=19.0 q=15.0 q=14.0 q=15.0 q=16.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=   [info] frame= 1086 fps= 20 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=13.0 q=16.0 q=15.0 q=14.0 q=14.0 q=12.0 q=13.0 q=13.0 q=13.0 q=13.0 q=12.0 size=   [info] frame= 1092 fps= 19 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=13.0 q=16.0 q=13.0 q=13.0 q=13.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 q=12.0 size=


  


  What I tried & My guesses

  


  My best guess is that the issue comes from one of the data buffers between the two applications (ffmpeg -> mpv). There are several buffers and I don't know exactly which ones, but there is at least a UDP buffer at the output of the server and another one at the input of the client.

  


  When a client is unreachable, the server's UDP buffer seems to fill up and thus don't continue streaming for other clients.

  


  I tried to tweak several parameters of ffmpeg concerning buffers but without success.

  


  

  • udp://100.64.0.32:1234?buffer_size=1024&connect=0&fifo_size=10&overrun_nonfatal=0

  


  • fps_mode

  


  • thread_queue_size

  


  


  Any help is welcome !