<html><head></head><body>Hello Marcello,<br><br>Verry good! But in my opinion the DNS server should not send out SOA queries faster than it can pickup the answers. Raising the buffers helps only up to a certain burst length. If you have more zones, the problem comes back.<br><br>Winfried<br><br><br><br><div class="gmail_quote">Am 29. Dezember 2020 09:17:06 MEZ schrieb Marcello Lupo <mlist@itspecialist.it>:<blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">
<pre class="k9mail">He Winfried,<br>yes it seems to be the same issue.<br>I tried to raise the kernel rmem limits with:<br><br>sysctl -w net.core.rmem_max=26214400<br>sysctl -w net.core.rmem_default=26214400<br><br>and the issue seems to be solved.<br>I will monitor it the next days.<br>Thank you<br>Bye<br>Marcello<br><br><blockquote class="gmail_quote" style="margin: 0pt 0pt 1ex 0.8ex; border-left: 1px solid #729fcf; padding-left: 1ex;">Il giorno 28 dic 2020, alle ore 22:26, Winfried Angele <winfried.angele@gmail.com> ha scritto:<br><br>Hi Marcello,<br><br>It looks like this is the same issue which I reported already:<br><br><a href="https://github.com/PowerDNS/pdns/issues/9112">https://github.com/PowerDNS/pdns/issues/9112</a><br><br>Maybe you can report your observations there as well.<br><br>Winfried<br><br><br>Am 28. Dezember 2020 21:41:21 MEZ schrieb Marcello Lupo via Pdns-users <pdns-users@mailman.powerdns.com>:<br>Hello,<br>I have configured a PDNS Authoritative server (v 4.4.0) with many slave zones provisioned with superslave and supermasters mechanism.<br>At the moment I have around 1400 zones provisioned from 5 bind master servers.<br><br>I’m getting a lot of:<br><br>Dec 28 21:00:02 pdns-db2 pdns_server-global[6838]: Received serial number updates for 374 zones, had 19 timeouts<br>Dec 28 21:05:23 pdns-db2 pdns_server-global[6838]: Received serial number updates for 395 zones, had 47 timeouts<br>Dec 28 21:06:34 pdns-db2 pdns_server-global[6838]: Received serial number updates for 303 zones, had 15 timeouts<br>Dec 28 21:16:07 pdns-db2 pdns_server-global[6838]: Received serial number updates for 394 zones, had 43 timeouts<br>Dec 28 21:21:25 pdns-db2 pdns_server-global[6838]: Received serial number updates for 599 zones, had 66 timeouts<br>Dec 28 21:22:41 pdns-db2 pdns_server-global[6838]: Received serial number updates for 521 zones, had 22 timeouts<br>Dec 28 21:26:56 pdns-db2 pdns_server-global[6838]: Received serial number updates for 455 zones, had 28 timeouts<br>Dec 28 21:28:08 pdns-db2 pdns_server-global[6838]: Received serial number updates for 235 zones, had 26 timeouts<br><br>I checked with tcpdump on both the master DNS and the powerdns itself and all answers to the AXFR requests starting from pdns server are correctly received on the pdns server but seems that are not handled by the system and give the timeout error. The udp-in-errors counter is raising every time of the exact number of the timeouts received. As the documentation says the udp-in-errors value seems to be related to packet received but cannot be handled from the pdns server itself.<br>I’m almost sure that it is a problem of resource tuning but I haven’t idea of what I should tune to avoid this behavior.<br>In the short future we will need to reach around 15k domains managed by this DNS server and I would like to do not have this issue.<br>In my scenario this server is dedicated only to the provisioning of domains and it do not answer to requests coming from internet. Requests coming from internet are handled from many servers balanced that rely on the mariadb replication of the PDNS DB main server.<br>Do you have any ideas?<br>Thank you<br>Bye<br>Marcello<br>Pdns-users mailing list<br>Pdns-users@mailman.powerdns.com<br><a href="https://mailman.powerdns.com/mailman/listinfo/pdns-users">https://mailman.powerdns.com/mailman/listinfo/pdns-users</a><br></blockquote><br></pre></blockquote></div></body></html>