I've had it reported 3 times in the last month, I've seen it twice. Not consistent at all. I just wish those types of errors good be logged so that a possible pattern or time of day could be found. Nothing in nginx logs, nothing in slow_query_log, nothing in php error log, or xenforo server logs.
I just wish those types of errors good be logged so that a possible pattern or time of day could be found. Nothing in nginx logs, nothing in slow_query_log, nothing in php error log, or xenforo server logs.
There really isn't anything that can be logged - it's a client side error. It sent the request off but didn't get a response in time (30 seconds). If there's no web server log that corresponds to this, then the request likely never got there, which would indicate a connectivity problem for that person.
It could also be certain services that XenForo uses that require connecting. For example on my test board, I set the mail server IP to something stupid (18.104.22.168), and anytime XF would try to send mail I would end up with the "server failed to respond" message... because ultimately when it was trying to connect to "my" mail server at 22.214.171.124, it would wait longer than 30 seconds.
It's actually one of the reasons I made an addon to send mail on shutdown vs. in realtime (the end user doesn't care if there really is a mail server issue with the *actual* mail server... and nothing they can do about it even if there is). Plus it makes the pages load faster since rendering time isn't dependent on mail being sent.
Thanks for that, I had seen some postings regarding misconfigured mail and the error, but it appears mail problems would throw something into the log file. I might try switching to SMTP to deliver mail just to rule that out. I think there was a mention about SMTP possibly slowing things down a bit due to it being synchronous, but I'm not running a lot of traffic at the moment. It's just something that's been nagging me that I haven't been able to figure out.
No, I haven't had time to wrap it up into something installable yet... But have you tried using the normal "Default" mail transport (sends it locally instead of making a network connection for mail sending)?
Yes, that's what I'm doing now, just because it's simple, quick, and asynchronous. I'm using the -f parameters, and haven't noticed any problems. I have noticed some mail being marked as spam though, but that's just because of a bad SPF record, or potentially other reasons. I haven't decided if a "no-reply@reverseDNSname" is better than trying to use something more recognizable or not, but that's a different thread.
As an update, a lot of my problems with connection, wierd Chrome errors, probably 90% of the errors I saw were resolved when I moved off Cloud Flare.
From my Windows machine, I couldn't even traceroute to any site on the CloudFlare network, although most times I could browse the site without too many problems. CloudFlare told me that the problem was due to my ISP doing something with ICMP traffic. I could do traceroutes to sites on CloudFlare in Linux (UDP), but not Windows (ICMP).
Even when I instituted the CloudFlare beta of "page rules", I had the same problem. It wasn't until I changed the name servers from CloudFlare to my original name servers did the problem go away.
CloudFlare support admitted that it was having some problems with Xenforo installations, and that's when they tried to get me to try out the "page rules" route, which still didn't work.
Personally, I think the problem lies in CloudFlare's DNS or how traffic is routed, or perhaps the way cached content is retrieved, but since I moved off that network, I've really had a lot better reliability.