hidden gull
Telegram is intermittent.
Messages often get double-check immediately, but no “typing”, no bot reply, and nothing reaches the dashboard.
Sometimes it recovers briefly, then fails again.
What logs show:
getUpdates hangs for ~170–180s
Polling stall detected
Network request for 'getUpdates' failed!
polling runner restarts after stall
during/after stall:
sendMessage failed
message processing failed
sometimes sendVideo failed
Important detail:
openclaw status often still shows Telegram ON / OK
but real behavior is broken/intermittent
What I tried:
Forced IPv4 preference in /etc/gai.conf
Temporarily disabled IPv6 with sysctl
Restarted gateway multiple times
Rebooted the whole machine
Recovered from a separate broken runtime/update state
Results:
IPv4 changes did not fix it
reboot did not fix it
after clean reboot, the same polling stalls returned
Network note:
https://api.telegram.org is reachable (HTTP 200)
this does not look like full Telegram outage
looks specific to Telegram long polling from this host/network
Current hypothesis:
Telegram long polling (getUpdates) is getting stuck on this host/network, causing the bot to stop receiving inbound messages even though the channel still appears healthy.
Questions:
Any known regression in Telegram polling?
Any recommended config for:
channels.telegram.network.autoSelectFamily
channels.telegram.network.dnsResultOrder
channels.telegram.timeoutSeconds
channels.telegram.pollingStallThresholdMs
Any recommended way to switch this setup away from polling or better debug stuck getUpdates?
sour vale