Thanks for your reply. Unfortunately the situation with .local is still
bad. That domain is often queried on DNS root servers and not only for
SOA record, which would make this heuristic probably the reason.
Interesting root server statistics:
https://stats.dns.icann.org/stats/d/wom-ext-5minagg-qtype-vs-tld/qtype-vs-tld?orgId=1&var-Server=IMRS&var-Region=All&var-Country=All&var-City=All&var-Instance=All
I think the problem is inability to disable avahi just on *some*
networks, where .local unicast domain might be used for some
network-specific content. nss-mdns package is optional, but you can
either disable it always or enable it always. There is no way to
specify: just on this network disable mdns resolution. File
/etc/mdns.allow might configure manually whether .local is acceptable
for multicast resolution. It would be used always. Just like disabling
avahi-daemon. That is what .local SOA? query allows, disabling mdns on
networks where it gets positive response. We lack a good way to disable
that resolution just sometimes.
More below...
On 07. 12. 22 21:14, Esko Dijk wrote:
Hi Petr,
Besides the mDNS RFC there is also RFC 6761 that defines the special-use domain names,
one of which is ".local". Both RFCs were completed early 2013 so nearing 10
years now.
Implementations that ignore the special status of ".local" are not just
'ignoring mDNS' but they are ignoring the foundations of the domain name system.
The special-use domains are not available for other purposes than their special use, so
for ".local" it means use for mDNS only is intended. Section 22.1 defines the
formal requirements.
A DNS server that follows these relevant IETF specs shouldn't be normally
configurable for .local use , however an exception is allowed for test use and
special use.
Also a resolver library by default does not create DNS unicast queries for
.local :
3. Name resolution APIs and libraries SHOULD recognize these names
as special and SHOULD NOT send queries for these names to their
configured (unicast) caching DNS server(s).
The "SHOULD/SHOULD NOT" allows an exception for unspecified exception cases; I
believe this was to cater for older resolver code and for legacy/test networks.
So it seems clear that Avahi and other software (some was mentioned in [1] I think) needs
by default to treat the .local domains as "special" and not do any unicast name
resolutions for it.
If software needs to be configured specifically to disable unicast queries for
.local then this is the wrong way around! Doing a manual configuration action
to enable the unicast .local queries would be acceptable, for the
aforementioned legacy, hacking, and testing purposes.
There is a problem some people were using .local zone even before MDNS
were created. They should be told that their configuration is wrong, but
I doubt nss-mdns is a good place for it. I think avahi-daemon might
report warning in such case, just once per start. It used to work and it
seems some people still rely on this unregistered domain name. Anyway,
any change for them would cause regression. That is why I am looking for
some backward compatible way at least partially.
In the latter case, when unicast .local queries are enabled, could the queries
be sent in parallel? The first-one-then-the-other waiting approach seems
strange.
No, parallel queries cannot be done easily in nsswitch. Because mdns
module implements just mdns and dns is implemented by another module and
the order of modules used is fixed. One module cannot call the other
one. It would be nice if the order could be changed dynamically, but the
API does not allow that. This is also privacy related. We do not want to
leak all names used on local network to DNS server far on the internet.
That is why under normal circumstances we do just mdns and if not found,
we end there for *.local names. On the other hand, DNS resolution is
quick, unlike long timeout required for NXDOMAIN response on mdns.
Since the special use of .local for mDNS remains valid even when someone decides to
configure their DNS server with ".local". If someone really wants only unicast
.local queries to be sent out it's always possible to disable Avahi.
I agree multicast resolution should be preferred, but I think we want
per-network configuration of mdns resolution, just like NM can do with
systemd-resolved. Manual change works fine for device fixed in just one
network. But with laptops and mobile devices that won't work well.
I would like to extend avahi-daemon eventually to be able to configure
used interfaces dynamically. But that will not be a tiny change.
Regards
Esko
-----Original Message-----
From: avahi <avahi-boun...@lists.freedesktop.org> On Behalf Of Petr Menšík
Sent: Wednesday, December 7, 2022 17:38
To: Avahi ML <avahi@lists.freedesktop.org>
Cc: ago...@google.com; zdoh...@redhat.com
Subject: [avahi] Is it time to drop .local SOA heuristic?
Hello everyone.
We have spent almost whole day discussing options about fixing issue [1]
with non-working .local resolution. It turned out the responsible were
unicast name server, which contains .local zone with SOA record.
I think current implementation is not the best one. I believe Apple
product is doing similar checks, but with different results [2]. If they
find .local SOA record, then they first send query to DNS, then if not
found there try at least mdns. That seems to be ideal way to solve it.
DNS can respond fast that no such name exists, unlike its multicast
counterpart.
Unfortunately nss-mdns plugin just skips MDNS, but never retries it
after the DNS. I see there two problems:
- mdns should not implement dns queries itself, especially when parallel
IPv4+IPv6 queries occur.
- resolve plugin does not allow anything pass after that in current
fedora's configuration. So appending fallback mdns plugin AFTER dns
would not work, because nothing reaches there when systemd-resolved is
active.
There is a question, whether still in year 2022, almost 10 years after
MDNS RFC were published, exist deployed .local zones in unicast DNS with
a real data. Would anyone know? Do we still want to prevent .local zone
breakage in unicast DNS?
Our results of the best available solution were this:
- if .local SOA is detected, do MDNS query anyway
- if the host is not found on MDNS, return UNAVAIL as result to nss.
Allow continuing to next module, resolve or dns. We expect
/etc/nsswitch.conf contains something like hosts: ... mdns4_minimal
[NOTFOUND=return] dns
- set shorter timeout in this case, so the DNS response is returned in 1
or 2 seconds. Current 5 seconds does not seem acceptable.
- (optional) move .local SOA query to avahi-daemon, so it can be cached
for at least few seconds.
If .local SOA is not found, then return NOT_FOUND and stop further
resolution.
An alternative would be removing this test at all and do mdns queries
always. If we had network-specific configuration of mdns, like
systemd-resolved can do, this would not be necessary. You could just
disable mdns where .local unicast domain provides useful data.
What would you think about that plan? Does it sound reasonable or not? I
think it would make it still usable even on legacy networks. Without
dramatic regression. Yes, it would add increased delay to unicast
*.local names, but otherwise they would stay working.
Any comments welcome. I have tried to do draft [3] on nss-mdns plugin,
which keeps timeout unchanged, but at least allows DNS query after
unsuccessful MDNS query. If someone would like to test it, I would be
grateful.
Regards,
Petr
1. https://bugzilla.redhat.com/show_bug.cgi?id=2148500
2. https://github.com/lathiat/nss-mdns/issues/75
3. https://github.com/lathiat/nss-mdns/pull/84
--
Petr Menšík
Software Engineer, RHEL
Red Hat, http://www.redhat.com/
PGP: DFCF908DB7C87E8E529925BC4931CA5B6C9FC5CB
