> On Oct 13, 2021, at 1:07 AM, Stephane Bortzmeyer <[email protected]> wrote: > > On Tue, Oct 12, 2021 at 01:01:08PM -0400, > Matthew Pounsett <[email protected]> wrote > a message of 11 lines which said: > >>> This might be a known intermittent IPv6 routing issue with DNSviz, do >>> you see this problem for v4 and/or v6 ? >> >> That would show up as a non-answer over IPv6, rather than an apparent >> PMTU/EDNS problem. > > DNSviz (and similar tools) may wrongly diagnose a PMTU problem if > there are random losses.
That's true. > 1) Try with bufsize=4096. No answer, because of a random packet loss. > 2) Retry with bufsize=1024. Answer received, therefore it must be a > PMTU problem. Actually, it give it several tries before reducing PMTU. Here is the full order of diagnostic queries when a timeout occurs: # 1 - no change # 2 - change timeout to 2 seconds # 3 - no change # 4 - reduce udp max payload to 512; change timeout to 1 second # 5 - change timeout to 2 seconds # 6 - remove EDNS option (if any); change timeout to 1 second # 7 - remove EDNS option (if any) # 8 - remove EDNS option (if any) # 9 - remove EDNS option (if any) # 10 - clear DO flag; # 11 - disable EDNS # 12 - return (give up) (See https://github.com/dnsviz/dnsviz/blob/master/dnsviz/query.py#L1766) Casey _______________________________________________ dns-operations mailing list [email protected] https://lists.dns-oarc.net/mailman/listinfo/dns-operations
