We stopped reading 2,195 MCP servers for three days. Another team's log is how we found out

Описание и применение

HISTOR перестал читать 2 195 серверов после перехода на новый протокол MCP, но показатели не сигнализировали об ошибке. Сравнение с независимым логом выявило проблему, после чего обновлённый клиент корректно откатывался к старому методу initialize, восстановив чтение всех серверов.

Задача

HISTOR’s MCP reader stopped reading 2,195 servers after a protocol change, but dashboards showed no alerts.

Результат

The reader resumed reading all affected servers; protocol‑version statistics revealed 19 % of servers support the new 2026‑07‑28 revision.

Рабочий процесс

• 1. Detect missing reads via comparison with a second log (ScraperWatch).
• 2. Identify error patterns (HTTP 401/403, legacy error IDs, missing supportedVersions).
• 3. Update reader to fall back to initialize on any of those patterns.
• 4. Deploy fix (08:44:57 UTC 1 Nov).
• 5. Verify reads resume and success metrics recover.

Доказательства

HISTOR’s reader learned the new revision and started trying it first. We made the fallback strict on purpose.

Источник

The reader now follows the specification's rule as written. It falls back to initialize, once, when the first new‑style request is refused.

Источник

The fix went live on 1 November at 08:44:57 UTC. In the first 90 seconds, 171 of 172 remote reads succeeded, and 145 of those used an older protocol.

Источник