[Dnsmasq-discuss] [PATCH] Retry queries only after giving the upstream server some time to respond
dl6er at dl6er.de
Mon Apr 5 20:30:56 UTC 2021
To be even more precise:
On Mon, 2021-04-05 at 22:16 +0200, Dominik Derigs wrote:
> This is the issue I'm concerned about. Some clients send the same
> multiple times (they don't seem to have a local cache).
These clients don't even intend them as retries. Wireshark confirms
they send them as individual queries (they have different IDs). Later
retries (which really rarely happen) have the same ID - as they should.
So maybe the fix could be distinguishing retries from the same source
as identified by the same ID and new queries for the same type/domain
In the latter case, we may be safe to skip forwarding again because
this is not meant as a retry from the client? My understanding is that
we can use the same argument for other clients requesting the same
type/domain at the same time. As long as no client sends a "real" retry
(same ID), we should be safe waiting on the first forwarded to appear.
So like 2.83/2.84 behavior but with a possibility for the clients to
actually trigger re-forwarding.
More information about the Dnsmasq-discuss