<div id="reply-content">
<br>
</div>
<div id="5D663E501BF34D30A3A7B75475A0F19F"><div><br></div>-- <br>Kyohei Moriyama<div><br></div></div>
<p style="color: #A0A0A8;">日付：2012年12月7日金曜日、時刻：22:14、差出人：Leo Baltus：</p>
<blockquote type="cite" style="border-left-style:solid;border-width:1px;margin-left:0px;padding-left:10px;">
<div id="quoted-message-content"><div><div>Op 07/12/2012 om 12:38:27 +0100, schreef W.C.A. Wijngaards:</div><blockquote type="cite"><div><div>On 12/07/2012 12:20 PM, Leo Baltus wrote:</div><blockquote type="cite"><div><div>Op 07/12/2012 om 11:50:04 +0100, schreef W.C.A. Wijngaards:</div><blockquote type="cite"><div><div>-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1</div><div><br></div><div>Hi Leo,</div><div><br></div><div>On 12/07/2012 11:29 AM, Leo Baltus wrote:</div><blockquote type="cite"><div><div>Hi,</div><div><br></div><div>When I run unbound multi-threaded perf keeps reporting</div><div>'Packets lost', but when it runs single threaded it does not.</div><div><br></div><div>$ perf -d 10 -a 'www.omroep.nl IN A' 2a02:458:101:31::34 qps: </div><div>29262.9 qps: 31986.4 qps: 32167.5 qps: 32479.7 qps: 33064.3</div><div>qps: 32174.4 qps: 31408.1 qps: 35887.4 qps: 37471.3 qps:</div><div>35364.5 overall time: 10.0001 sec Packets lost: 732</div><div>0(NOERROR): 331249 replies average qps: 33124.5</div><div><br></div><div>This is unbound-1.4.18 with openssl-1.0.0j, ldns-1.6.16 and </div><div>libevent-2.0.21-stable running on a linux-2.6.27.59 kernel.</div><div><br></div><div>It even reports this when perf is run on the host running</div><div>unbound.</div><div><br></div><div>Is this a known issue?</div></div></blockquote><div><br></div><div>No. Can you try setting so-rcvbuf higher (4m) ? It may be lost </div><div>simply because it is out of buffers. There is nothing intrinsic</div><div>in unbound that makes it drop packets in this setup; it must be</div><div>the OS.</div></div></blockquote><div><br></div><div>It makes no difference.</div><div><br></div><div>here's the multithreading part of my unbound.conf:</div><div><br></div><div>num-threads: 4 outgoing-range: 8192 num-queries-per-thread: 4096 </div><div>so-rcvbuf: 4m</div><div><br></div><div>Also 'netstat -su' does not report any increments on 'receive</div><div>errors' or RcvbufErrors.</div></div></blockquote><div><br></div><div>Perhaps perf fails to count them? They may still be in transit, and</div><div>perf stops abruptly, counts them as lost? tcpdump may be able to</div><div>count the number of packets that transited and where those 732 packets</div><div>went?</div></div></blockquote><div><br></div><div>I think you are right.</div><div><br></div><div>Once perf stopped receiving, the client started to send 'ICMP dest.</div><div>unreachable'. If I add the number of packets, they get pretty close to the</div><div>number of packets lost reported by perf.</div><div><br></div><div>So maybe perf should just stop sending after $duration (or SIGHUP) and</div><div>wait a while for the last packets to arrive?</div><div><br></div><div>-- </div><div>Leo Baltus, internetbeheerder /\</div><div>NPO ICT Internet Services /NPO/\</div><div>Sumatralaan 45, 1217 GP Hilversum, Filmcentrum, west \ /\/</div><div>servicedesk@omroep.nl, 035-6773555 \/</div><div>_______________________________________________</div><div>Unbound-users mailing list</div><div>Unbound-users@unbound.net</div><div>http://unbound.nlnetlabs.nl/mailman/listinfo/unbound-users</div></div></div>
</blockquote>
<div>
<br>
</div>