Re: Archives down? 503 error

* Magnus Hagander (magnus(at)hagander(dot)net) wrote:
> On Oct 6, 2015 4:36 AM, "Stephen Frost" <sfrost(at)snowman(dot)net> wrote:
> >
> > * Bruce Momjian (bruce(at)momjian(dot)us) wrote:
> > > I am seeing a 503 error from our archives:
> > >
> > >
> http://www.postgresql.org/message-id/54B70C3F.4020102@dunslane.net
> > >
> > > Error 503 Backend fetch failed
> > >
> > > Backend fetch failed
> > > Guru Meditation:
> > >
> > > XID: 13756651
> > >
> > > Varnish cache server
> >
> > Yeah, zalkon was having issues resolving DNS for wrigleys. Seemed to be
> > some kind of intermittent ipv6 or possibly ipv6+DNS issue. Also appears
> > to have now cleared by itself..
>
> Just to be clear, neither DNS or wrigleys have anything at all to do with
> delivering the archives Web pages. And I'm pretty sure the backend
> communication is actually ally v4 on that one.

Err, right, the archives are off of arkaria, the main website backend is
wrigleys. The first issue that I noticed was that DNS wasn't working
and while I was trying to figure out what was going on, things seemed to
come back.

> That said, general network issues obviously does, so that's probably what
> the issue was.

Yeah. They were rather odd issues- borg couldn't make TCP/53
connections to secarus or other systems, triona was failing DNS for
things in postgresql.org, and borg also couldn't get to some hosts for a
period of time. Further, some systems were getting connection refused
errors when trying to hit mahout on 443.

Seemed to be more than the usual run-of-the-mill network connectivity
issues as we were getting actual connection refused responses in some
caes like someone's IDS went insane or a firewall popped up in front of
us somewhere.

In any case, things did end up resolving themselves relatively quickly..

Thanks!

Stephen