Replies: 5 comments 1 reply
-
Gonka Node Observability PlatformProposal от INC4 | 16 Апрель 2026 Запрос финансирования: $96,000 USD (USDT) на 12 месяцев Содержание
1. РезюмеСуществующие эксплореры и дашборды показывают только on-chain данные, оставляя off-chain состояние валидаторов полностью непрозрачным. Немногие операторы, которые всё же мониторят свою инфраструктуру, используют разные инструменты, разные метрики и разные baseline — что приводит к разным интерпретациям и затрудняет координацию при возникновении проблем. У сети нет единого источника правды и общего фреймворка для оценки здоровья валидаторов. Даже сейчас, просто наблюдая за работой валидаторов, можно заметить, что многие не реагируют на технические проблемы вовремя — растущий Inference Miss Rate, падающий CPoC Ratio, отставание Network Node Sync. Эти паттерны сохраняются, потому что у операторов нет возможности увидеть ранние предупреждающие сигналы или получить экстренные алерты внутри собственной инфраструктуры. Для решения этой проблемы INC4 — активный оператор валидаторов Gonka с практическим опытом в блокчейн-инфраструктуре и наблюдаемости — предлагает создать Gonka Node Observability Platform — open-source стек мониторинга с добровольным участием, который агрегирует off-chain метрики от Network-нод и ML-нод в общий, публично доступный дашборд. Платформа разворачивается на независимой облачной инфраструктуре — без использования ресурсов конкретных валидаторов и без предоставления кому-либо привилегированного доступа — чтобы все операторы имели равный доступ к данным и равную видимость. Для Core Team — единое представление всей сети, видимость проблемных нод и временных периодов, данные для обоснованных решений по протоколу, SLA-отчёты и возможность оценить масштаб сетевых проблем до и после обновлений. Для Отдельных Валидаторов — алерты в реальном времени, сравнение производительности со средними по сети, добровольный обмен логами для помощи в диагностике и интерпретации метрик, и не нужно строить собственный стек мониторинга. Обнаружение и предотвращение даже одного крупного общесетевого инцидента или серии более мелких инцидентов у отдельных валидаторов может сэкономить экосистеме значительно больше, чем годовая стоимость этой платформы. Главными выгодополучателями станут сами операторы валидаторов, которые несут прямые издержки от каждой пропущенной эпохи и каждого часа недиагностированного простоя — особенно индивидуальные хосты без выделенных DevOps-специалистов, для которых создание и поддержка сопоставимого стека мониторинга своими силами попросту нереализуемы. Мы запрашиваем $96,000 в USDT на 12 месяцев с выплатой квартальными траншами для развёртывания и поддержки production-grade платформы — кастомные Gonka-экспортёры, общесетевые и индивидуальные дашборды, добровольная агрегация логов, внешние проверки доступности эндпоинтов, алертинг и SLA-отчётность, практический онбординг валидаторов, поддержка реагирования на инциденты и постоянная операционная поддержка. Весь код, конфигурации и дашборды будут open-source и опубликованы в публичных GitHub-репозиториях. 2. Описание проблемыOff-chain состояние сети не видноGonka — растущая сеть с более чем сотней валидаторов, ещё большим количеством ML-нод и суммарным парком GPU, превышающим 3,000 карт. Существующие блокчейн-эксплореры и дашборды показывают on-chain данные — высоту блоков, транзакции, voting power. Но нет ни одного инструмента для наблюдения за off-chain состоянием сети: здоровье GPU, статус контейнеров, корневые причины miss rate, время загрузки моделей, метрики производительности LLM, тренды инфраструктуры и т.д. Каждый оператор мониторит свою инфраструктуру в изоляции — или не мониторит вообще. Те, кто мониторят, настраивают собственные инструменты, считают метрики по-разному и используют разные baseline — что приводит к недопониманию и путанице при обсуждении сетевых проблем. В частности, существующие инструменты не показывают:
На практике валидаторы сталкивались с длительными периодами высокого miss rate или простоя инференса, не имея возможности определить причину. При этом у Core Team не было способа увидеть масштаб таких проблем по всей сети. Такие ситуации приводят к потере наград для операторов и замедленной реакции команды — проблемы, которые общая платформа мониторинга помогла бы обнаружить и решить значительно быстрее. Что происходит без решения
3. Контекст индустрииСбор метрик с валидаторов в единое место — не новая идеяАгрегация телеметрии и метрик от валидаторов в общий backend — устоявшаяся практика в блокчейн-индустрии. Множество крупных сетей уже это делают:
Это не экзотическая идея. Именно так зрелые блокчейны получают видимость здоровья своей сети, быстрее диагностируют проблемы и принимают решения об обновлениях протокола на основе данных. Во многих блокчейн-сетях телеметрия с нод собирается без полного ведома операторов — телеметрия часто включена по умолчанию в софте ноды, а в некоторых случаях у операторов вообще нет возможности её отключить. В отличие от этого, Gonka Node Observability Platform спроектирована как полностью opt-in система — валидаторы сами решают, участвовать ли, и никакие данные не собираются без их явного действия. Чем больше валидаторов подключится, тем точнее и полнее будет картина здоровья сети. Платформа с 30% подключённых валидаторов даёт полезные инсайты; платформа с 80% становится надёжным источником правды для всей экосистемы. 4. Предлагаемое решениеGonka Node Observability PlatformManaged open-source стек мониторинга, куда операторы валидаторов добровольно отправляют off-chain метрики на общую платформу, поддерживаемую INC4. Принципы дизайна
Создаваемая ценностьДля Core Team:
Для валидаторов:
Для сообщества:
5. Технический подходПлатформа будет развёрнута на распределённой облачной инфраструктуре, что обеспечивает:
Мы будем использовать хорошо зарекомендовавшие себя, проверенные индустрией инструменты для мониторинга: Prometheus для сбора метрик, Grafana для дашбордов и визуализации, Alertmanager для уведомлений, Promtail/Loki для унифицированной добровольной агрегации логов и PagerDuty для управления инцидентами и эскалации дежурств. INC4 имеет практический опыт создания и эксплуатации инфраструктуры мониторинга для блокчейн-сетей. Выбор каждого компонента стека продиктован реальными операционными требованиями — надёжность под нагрузкой, простота интеграции с существующими настройками валидаторов, минимальные ресурсные затраты на стороне ноды и возможность масштабирования без перестройки архитектуры по мере роста сети. Этот практический опыт напрямую определяет архитектурные и инструментальные решения, лежащие в основе данной платформы. Детальная архитектура, включая конкретные определения метрик, потоки данных и спецификации экспортёров, будет задокументирована отдельно и будет эволюционировать по мере развития платформы. 6. Scope и Deliverables
7. Бюджет и график выплатСводка
График выплат
Каждый транш выплачивается первого числа соответствующего периода. Первый транш больше, так как покрывает развёртывание инфраструктуры и наиболее интенсивную фазу разработки проекта. Риски
Бюджет рассчитан на один год. После первого года условия могут быть пересмотрены и продлены на тех же или скорректированных условиях. INC4 опубликует прозрачный отчёт об использовании платформы, adoption и расходах по итогам грантового периода — давая сообществу чёткую основу для решения о продлении. Если сообщество решит не продлевать, полностью настроенная и работающая платформа — включая всю инфраструктуру, код и конфигурации — может быть передана сообществу или Core Team. 8. Критерии успехаЧто поставляет INC4:
Что зависит от сообщества:
Ключевые показатели эффективности:
9. Команда
INC4 — активный участник экосистемы Gonka. Мы эксплуатируем валидаторы на mainnet и testnet, разрабатываем приложения для сети Gonka. Этот proposal вырос из нашего прямого опыта — мы сами столкнулись с отсутствием общесетевой видимости как операторы валидаторов и хотим решить эту проблему для всей сети. INC4 вовлечён в несколько инициатив в экосистеме Gonka — платформа наблюдаемости лишь одна из них. Например, мы также разрабатываем NOP (Node Onboarding Package) — open-source утилиту для быстрого запуска валидаторов (https://github.com/inc4/gonka-nop). Наше участие в сети — долгосрочное и не ограничивается данным proposal. Как компания, INC4 основана в 2013 году, 70+ инженеров и 230+ реализованных проектов в блокчейн-инфраструктуре и AI-системах. Практический опыт в создании и поддержке майнинг-инфраструктуры для Bitcoin, Ethereum, Filecoin. |
Beta Was this translation helpful? Give feedback.
-
|
Hi guys, I’ve had experience working with you before, and in any case I sincerely wish you success and hope you remain with Gonka. May I kindly ask whether it would be possible to cancel any of the future tranches if plans were to change later on? |
Beta Was this translation helpful? Give feedback.
-
|
Vesting Contract: https://github.com/rwxr-xr-x/gonka-usdt-vesting-schedule |
Beta Was this translation helpful? Give feedback.
-
|
Questions according to proposal:
It is very high-level view on the proposal and it seams to lack a lot of technical details that should be covered in the spec first before it could be applyed. Moreover it would be much better to see first proof of concept |
Beta Was this translation helpful? Give feedback.
-
|
This is a cover letter for a funding request. It typically includes a high-level overview of problem statement and proposed solution, without too much detailed information on the budget or technical implementation. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Gonka Node Observability Platform
Proposal by INC4 (https://inc4.net) | 16 April 2026
Funding Request: $96,000 USD (USDT) for 12 months
Table of Contents
1. Executive Summary
Today's explorers and dashboards only show on-chain data, leaving the off-chain state of validators completely opaque. The few operators who do run their own monitoring use different tools, different metrics, and different baselines, leading to different interpretations and making it harder to coordinate when problems arise. The network lacks a single source of truth and a common framework for measuring validator health.
Even today, simply watching how validators operate reveals that many are not reacting to technical problems in time — growing Inference Miss Rates, dropping CPoC Ratios, network sync falling behind. These patterns persist because operators have no way to see the early warning signs or get emergency alerts inside their own infrastructure.
To address this, INC4 — an active Gonka validator operator with hands-on experience in blockchain infrastructure and observability — proposes the creation of the Gonka Node Observability Platform — an open-source, opt-in observability stack that aggregates off-chain metrics from Network nodes and ML nodes into a shared, publicly accessible dashboard. The platform is deployed on independent cloud infrastructure — without using resources of any individual validator or granting anyone privileged access — so that all operators have equal access and visibility into the data.
For the Core Team — a unified view of the entire network, visibility into problematic nodes and time periods, data for informed protocol decisions, SLA reports, and the ability to assess the scope of network-wide issues before and after upgrades. For Individual Validators — real-time alerts, performance comparison against network averages, opt-in log sharing to get help with troubleshooting and metrics interpretation, and no need to build your own monitoring stack.
Detecting and preventing even a single major network-wide incident, or a series of smaller incidents at individual validators, can save the ecosystem far more than the annual cost of this platform. The primary beneficiaries are the validator operators themselves, who bear the direct cost of every missed epoch and every hour of undiagnosed downtime — especially individual hosts without dedicated DevOps staff, for whom building and maintaining a comparable monitoring stack on their own is simply not feasible.
We request $96,000 in USDT over 12 months, paid in quarterly tranches, to deploy and maintain a production-grade observability platform — custom Gonka exporters, fleet-wide and individual dashboards, opt-in log aggregation, external endpoint health checks, alerting and SLA reporting, hands-on validator onboarding, incident response support, and ongoing operational maintenance. All code, configurations, and dashboards will be open-source and published in public GitHub repositories.
2. Problem Statement
Off-chain state of the network is not visible
Gonka is a growing network with over a hundred validators, an even larger number of ML nodes, and a combined GPU fleet exceeding 3,000 cards. Existing block explorers and dashboards show on-chain data — block heights, transactions, voting power. But there are zero tools for network-wide off-chain observability — GPU health, container status, miss rate root causes, model load times, LLM performance metrics, infrastructure trends, etc. Each operator monitors their infrastructure in isolation — or doesn't monitor at all. Those who do monitor set up their own tools, calculate metrics differently, and use different baselines — which leads to miscommunication and confusion when discussing network issues.
Specifically, existing tools do not show:
In practice, validators have experienced prolonged periods of high miss rate or inference downtime without being able to identify the root cause. At the same time, the Core Team had no way to see the scope of such issues across the network. These situations result in lost rewards for operators and delayed response from the team — problems that a shared observability platform would help detect and resolve much faster.
What happens without a solution
3. Industry Context
Collecting validator metrics in one place is not new
Aggregating telemetry and metrics from validators into a shared backend is a well-established practice across the blockchain industry. Multiple major networks already do this:
https://metrics.solana.com:3000https://telemetry.polkadot.iohttps://telemetry.polkadot.io/#list/Kusamatelemetry.nearone.org) and pushes data every 10 secondstelemetry.mainnet.aptoslabs.com) by default; the architecture is documented in a public SPECotel.celestia.observer) for DA nodes, plus a Prometheus-based observability stack for consensus nodesThis is not an exotic idea. It is how mature networks gain visibility into their health, diagnose issues faster, and make data-driven protocol decisions.
In many blockchain networks, node telemetry is collected without operators being fully aware of it — telemetry is often enabled by default in the node software, and in some cases operators have no way to disable it at all.
By contrast, the Gonka Node Observability Platform is designed as a fully opt-in system — validators choose to participate, and no data is collected without their explicit action.
The more validators that join, the more accurate and complete the picture of network health becomes. A platform with 30% of validators connected provides useful insights; one with 80% becomes a reliable source of truth for the entire ecosystem.
4. Proposed Solution
Gonka Node Observability Platform
A managed, open-source observability stack where validator operators voluntarily push off-chain metrics to a shared platform maintained by INC4.
Design principles
Value delivered
For the Core Team:
For validators:
For the community:
5. Technical Approach
The platform will be deployed on distributed cloud infrastructure, providing:
We will use well-established, industry-proven tools for observability: Prometheus for metrics collection, Grafana for dashboards and visualization, Alertmanager for notifications, Promtail/Loki for unified opt-in log aggregation, and PagerDuty for incident management and on-call escalation.
INC4 has hands-on experience building and operating observability infrastructure for blockchain networks. The choice of each component in the stack is driven by real-world operational requirements — reliability under load, ease of integration with existing validator setups, minimal resource overhead on the node side, and the ability to scale without rearchitecting as the network grows. This practical experience directly informs the architecture and tooling choices behind this platform.
The detailed architecture, including specific metric definitions, data flows, and exporter specifications, will be documented separately and will evolve as the platform matures.
6. Scope and Deliverables
7. Budget and Payment Schedule
Summary
Payment schedule
Vesting contact: https://github.com/rwxr-xr-x/gonka-usdt-vesting-schedule
Each tranche is paid on the first day of the respective period.
The first tranche is larger because it covers the infrastructure setup and the most development-intensive phase of the project.
Risks
The budget is calculated for one year. After the first year, the arrangement can be reviewed and renewed on the same or adjusted terms. INC4 will publish a transparent report on platform usage, adoption, and costs at the end of the grant period — giving the community a clear basis for the renewal decision. If the community decides not to renew, the fully configured and operational platform — including all infrastructure, code, and configurations — may be transferred to the Core Team.
8. Success Criteria
What INC4 delivers:
What depends on the community:
Key Performance Indicators:
9. Team
INC4 is an active participant in the Gonka ecosystem. We operate validators on mainnet and testnet, and develop applications for the Gonka network. This proposal grows out of our direct experience — we face the lack of network-wide visibility firsthand as validator operators and want to solve this problem for the entire network.
INC4 is involved in multiple initiatives across the Gonka ecosystem — the observability platform is one of them. For example, we also develop NOP (Node Onboarding Package) — an open-source utility for fast validator deployment (https://github.com/inc4/gonka-nop). Our commitment to the network is long-term and not limited to this proposal.
As a company, INC4 was founded in 2013, with 70+ engineers and 230+ delivered projects in blockchain infrastructure and AI systems. Hands-on experience in building and maintaining mining infrastructure for Bitcoin, Ethereum, Filecoin.
Beta Was this translation helpful? Give feedback.
All reactions