[leafnode-list] Re: Handling fetchnews failures in a more friendly fashion

[Matthias Andree]

Am 10.03.23 um 19:26 schrieb Robert Marshall via leafnode-list:
> I'm using leafnode 2 and using eternal-september and gmane.io as upstreams
> Recently eternal-september has been a little unreliable (at least for me!)
>
> news.eternal-september.org: connecting to port nntp
>    trying:    address 88.198.57.247 port 119...
>    connected: address 88.198.57.247 port 119.
> news.eternal-september.org NNTP server disconnected or timed out while waiting for response
> news.eternal-september.org: Server didn't want to talk to us, reply code -1
> ERROR: FETCHNEWS MUST REFETCH THE WHOLE ACTIVE FILE NEXT RUN.
> REASON: could not successfully talk to all servers.
>
> And at that point the local clients can't connect to the news spool on
> the local server. Is there any way of preventing the clients from
> stopping working until eternal-september wants to talk to me again?
>
>
> Robert

Hi Robert,

do you happen to have any leafnode-side debug logging

a. of the fetchnews failure,

b. of the "local clients can't connect"... issue?
What are the protocol errors the clients see, and what is the logging?


The situation you see should only strike if news.eternal-september.org
flaked out while fetchnews was refreshing the active file (= newsgroups
list) from said server, so this would not normally be a common
situation, but feel free to report details if your log has evidence to
the contrary.