[Pdns-users] Occasional TCP query timeouts.

[Grant Keller]

Here are the pcaps from the server in question.

ftp://ftp.sonic.net/pub/users/gkeller/tcp_pcap

Thank you for your help so far.


> Hello Grant,
>
> On Sep 30, 2011, at 7:12 PM, Grant Keller wrote:
>
>> We have been having this problem with TCP queries timing out on one of our nameservers running pdns 3. We have a monitor that runs every two minutes, and every half hour or so it times out on its query. I can reproduce this by  running the following:
>>
>> while [ 1 ]; do
>> dig schwer.us +recurse +tcp +short +time=1 +retry=1 @c.ns.sr.sonic.net
>> date
>> sleep 1
>> done
>>
>> If I let that run for a while, it will capture several timeouts:
>>
>> ;<<>>  DiG 9.7.3<<>>  schwer.us +recurse +tcp +short +time=1 +retry=1 @c.ns.sr.sonic.net
>> ;; global options: +cmd
>> ;; connection timed out; no servers could be reached
>>
>> The pdns.log has no significant entries at the times that this occures. I have watched iptraf as I recieve the timeouts, and It dosn't seem that we are hitting our max number of tcp connections (1000). This problem first started in pdns-2.9.22, and continued after the upgrade. None of our other name servers have this problem.
>>
>> What would be the next step in troubleshooting this? Are there any extra logging options I can turn on? Thanks in advance.
>
> Most important to know, of course, is whether the queries even reach pdns. Are you in a position to run tcpdump/tshark on c.ns.sr.sonic.net, to monitor requests from a test run like the one described above?
>
> Kind regards,
> Peter van Dijk



-- 
Grant Keller