feat(client): add watchdog and disconnect flap handling
Introduce an active watchdog that probes the Telegram client to detect zombie sessions and restart them in‑process. Add configurable disconnect flap detection with a sliding window to trigger restarts after repeated disconnects. New environment variables and config entries are added, and the Kurigram dependency is now version‑pinned.
This commit is contained in:
+9
-1
@@ -8,10 +8,18 @@ services:
|
||||
environment:
|
||||
TG_API_ID: XXX
|
||||
TG_API_HASH: XXX
|
||||
# TG_PROXY_HOST: 127.0.0.1 # MTProto proxy host, SOCKS5 interface (optional)
|
||||
# TG_PROXY_HOST: 10.0.0.1 # MTProto proxy host (SOCKS5). Prefer a literal IP over a hostname:
|
||||
# # a hostname is re-resolved on every reconnect, so flaky DNS during a
|
||||
# # network blip can wedge Pyrogram's reconnect.
|
||||
# TG_PROXY_PORT: 1080 # MTProto proxy SOCKS5 port (default: 1080)
|
||||
# TG_PROXY_USERNAME: XXX # SOCKS5 username (optional)
|
||||
# TG_PROXY_PASSWORD: XXX # SOCKS5 password (optional)
|
||||
# TG_WATCHDOG_ENABLED: "True" # Active liveness watchdog: detects 'zombie' sessions and restarts in-process (default: True)
|
||||
# TG_WATCHDOG_INTERVAL: 60 # Seconds between liveness probes (default: 60)
|
||||
# TG_WATCHDOG_TIMEOUT: 10 # Seconds to wait for each get_me probe (default: 10)
|
||||
# TG_WATCHDOG_FAILURES: 3 # Consecutive failed probes before restart (default: 3)
|
||||
# TG_DISCONNECT_FLAP_LIMIT: 3 # Disconnect events within the flap window before an in-process restart (default: 3)
|
||||
# TG_DISCONNECT_FLAP_WINDOW: 120 # Flap detection window in seconds (default: 120)
|
||||
PYROGRAM_BRIDGE_URL: https://pgbridge.example.com
|
||||
API_PORT: 80
|
||||
TOKEN: ХХХ
|
||||
|
||||
Reference in New Issue
Block a user