[Pdns-users] recursor stops answering under stess

[Juergen Georgi]

Hi,

while doing perfomance testing on an Sun-Fire-V240 (Sparc, Solaris-10),
pdns_recursor 3.1.4 just stopped answering on its bge0 interface.
However, it was still responding to queries on 127.0.0.1. 

The testing was done with the "queryperf" tool from the bind-9
distribution, called locally with

queryperf -q 200 -t 5 -s <ip> -d queries

File "queries" contained ca. 680.000 original queries
collected on a production dns server running bind-9.

The logfile showed just the following entries:

Dec 13 16:46:41 ... [ID 702911 daemon.error] stats: 442366 questions, 273871 cache entries, 171144 negative entries, 0% cache hits
Dec 13 16:46:41 ... [ID 702911 daemon.error] stats: throttle map: 18566, ns speeds: 21643
Dec 13 16:46:41 ... [ID 702911 daemon.error] stats: outpacket/query ratio 277%, 0% throttled, 0 no-delegation drops
Dec 13 16:46:41 ... [ID 702911 daemon.error] stats: 2304 outgoing tcp connections, 204 queries running, 133790 outgoing timeouts

The silence on bge0 occurred at about 500.000 queries 
and continued even after shutdown of queryperf.

Could this be the effect of throttling - a client punishment
somehow? 

"5 timeouts in 20 seconds also lead to query suppression"

comes to my mind. I couldn't query the server from a different 
client ip, since the box was barred by ipf.

Today I repeated the test run in order to reproduce the
failure, but id did not happen again.

Best regards,

-Juergen Georgi