I have DNSdist balancing ~14k QPS to four PDNSr servers. DNSdist queries the backend servers via TCP. On all four PDNSr servers, I'm seeing both of the following log messages:
pdns-recursor[4561]: msg="Timeout from remote TCP client" subsystem="in" level="0" prio="Warning" tid="4" ts="1720021555.321" proto="tcp" remote="<DNSdistIP>:44808"
pdns-recursor[4561]: msg="Unable to parse packet from TCP client" subsystem="in" level="0" prio="Error" tid="1" ts="1720022346.840" proto="tcp" remte="<DNSdistIP>:49766"
According to the DNSdist and PDNSr web GUIs and lack of complaints from subscribers, queries are being answered, no drops.
I'm working to tune this system to better handle a larger volume (including bursts) of queries - I'm stuck with the latter logs. Any insight on how to fix this is greatly appreciated.
