DNS Resolvers seem to have burped (?) Options

[hulkster]

So I noticed a number of things running slow on a couple of long-time servers. Load factor was minimal and no major changes on the machines. In the past, this has been indicative of DNS resolving problems.

They are located in Dllstx4 and I confirmed I'm using 67.19.{0,1}.10 in /etc/resolv.conf per Tony's post

However, if I do an "dig www.cnn.com", I get
; <<>> DiG 9.2.4 <<>> www.cnn.com
;; global options: printcmd
;; connection timed out; no servers could be reached
I try this a couple of times, and I'll occasionally get a response ... but not the nornal snappy return.

Sounds a lot like this previously reported problem which seems to crop up every couple of months.

Please also see Ticket #4257416PLNT for a previous instance of this (with resolution from Planet'ers) and I have filed ticket 5113370PLNT for the current problem.

Thanks in advance for looking into and suspect there must be some root cause issue here ...
alek

[Jeff]

Seeing the same thing here
dig cnn.com

; <<>> DiG 9.3.4-P1 <<>> cnn.com
;; global options: printcmd
;; connection timed out; no servers could be reached

Added a resolver from one of the other datacenters as a temporary quick fix.

[hulkster]

Quick response on the ticket ... with advice being to use a name server from Level 3, AT&T, etc.

While that works, shouldn't we be able to use the Planet's nameservers ... and what was up with the 67.19.{0,1}.10 machines ... again, this problem crops up every once in a while.

[Jeff]

Are the resolvers listed in resolv.conf used in order or randomly?

I never really thought about using out-of-datacenter dns servers due to slightly increased latency, but it would be nice if we could specify them as failover resolvers, though I would have thought that using one of the other pairs would be faster than going all the way to L3/ATT resolvers...