> On Oct 5, 2021, at 5:34 AM, Jürgen Botz via Outages <[email protected]> > wrote: > > OK, aside from the Facebook outage, and no doubt because of it, we > effectively had an outages outage... messages to the lsit were massively > delayed, causing lots of duplicate reports and general failure to have > any meaningful confirmations, etc. > > Why? What happened? Well, outages is hosted on a server > (puck.nether.net) with a couple dozen other lists, and some of these lists > probably have members with email addresses at facebook domains, > and the SMTP server kept trying to look those up and having wait for > timeouts? Something like that, except that since there were no routes > to facebook's nameservers there shouldn't have been any need to wait > for timeouts. Anyone here form nether.net actually take a look? > > It's more than idle curiosity, I'd like make my mail servers more > resilient to this kind of situation, and the outages list of all > probably should be, too.
Sure, I’ve gone and tweaked a few more things, there’s a balancing act here between having a single system send out lots of concurrent mail to a server, eg: 60756540E50* 22760 Tue Oct 5 01:28:00 [email protected] (host itchy.cerento.com[199.190.154.20] said: 451 Only one recipient at a time (in reply to RCPT TO command)) [email protected] Some systems are explicitly configured to not be optimized (see above, then imagine how many people in parallel @gmail might get a message) And there are many people with old subscriptions or domains that don’t exist anymore which gets to be exciting when everything is going on at once: (connect to mail.bestii.com[172.96.180.81]:25: Connection timed out) [email protected] (connect to canadawebhosting.com[74.201.58.138]:25: Connection timed out) [email protected] (connect to muhpanel.ironusmaidenus.com[51.15.246.204]:25: Connection refused) [email protected] (connect to mail.mailhost4.com[198.50.245.163]:25: Connection refused) [email protected] (connect to mail.etsms.com[50.232.238.69]:25: Connection timed out) [email protected] (connect to keyedupmedia.com[204.11.56.48]:25: Connection timed out) [email protected] (connect to metacloud.com[72.52.10.14]:25: Connection timed out) [email protected] (connect to smtp.naturalwireless.com[52.119.91.194]:25: Connection timed out) [email protected] [email protected] (host s1.mail.pciwest.net[2604:2400:a::425] said: 450 4.7.1 Bad Attachment .com (in reply to end of DATA command)) [email protected] (delivery temporarily suspended: connect to mail.talueee.com[159.69.230.243]:25: No route to host) [email protected] (connect to mail-1.meridian-enviro.com[198.58.69.47]:25: Connection timed out) [email protected] (reason unavailable) (All of these are from the same message btw) I was away for personal reasons yesterday so missed out on all the fun. I’ll check how this message goes out and see if the tweaks helped. - Jared _______________________________________________ Outages mailing list [email protected] https://puck.nether.net/mailman/listinfo/outages
