Hello everyone, We are seeing very vague issues with our PowerDNS setup where certain sequences of requests can cause full queues and dropped queries. Under normal circumstances, the server can handle more than 10 kqueries/sec, but when the bug is triggered, performance drops to 300-400 queries/sec.
There seems to be a correlation with the amount of domains that are queried which return REFUSED. DNSSEC makes it easier to trigger this problem, but disabling it does not make it disappear. It is reproducible with both MySQL and PostgreSQL used as a backend, so it doesn't seem to be an issue with the database. The schemas and indices are correct. Our conclusions so far are that there is some kind of bottleneck between PowerDNS and the database, but either are fast enough under normal operation. The slow requests are cached, because a second run within the TTL does not cause slowness. We have already tried tuning these variables, and while it does seem to help, it only masks the real issue. We have run out of ideas where to look and what to tune. Perhaps anyone here could help us further? -- Regards, Martijn Reening _______________________________________________ Pdns-users mailing list Pdns-users@mailman.powerdns.com https://mailman.powerdns.com/mailman/listinfo/pdns-users
