Development: DNS + Postfix Outage

DNS Outage
---

While upgrading the l10n.gnome.org machine, I hit a few issues while 
upgrading the bind9 package that was failing to restart for an openssl 
error. (which seems to be known, [1])

The same machine hosted our secondary DNS server but that wasn't a 
problem since our main host (clipboard) was actually up and running. 
Suddenly all the DNS zones got dropped and all GNOME services were 
actually unreachable. 

After a wider look at the Whois informations, we actually found out 
our old main DNS server (menubar) was still being listed as a 
nameserver, but no DNS services were hosted there anymore. (so 0 out 
of 2 DNS services were active)

We finally updated the zones information on menubar and restarted the 
DNS service while we actually wait the relevant folks to update Whois 
details with our new main DNS host.

Postfix Outage
---

As you may have noticed all the < at >gnome.org mails weren't being 
delivered correctly (bugzilla, mailing lists, personal aliases etc.).

For some reason Amavisd was requesting Postix's data not through 
the usual localhost address (127.0.0.1) but using its public IP 
instead. (that will obviously fail since access is restricted to 
localhost only)

After looking at the logs et all I found out telnet wasn't working 
correctly when trying to connect to the Amavisd port on localhost. 
Adding a new Iptables rule fixed the problem, the queue is currently 
flushing so expect to receive a good bunch of GNOME mails in the near 
future.

Thanks for your patience and understanding,

[1] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=668952

Comments are closed.