oryxlabs / polardns Goto Github PK

as of python 3.11 toml parser is part of standard library.

what does it mean for the project is as follow.

• remove third party dependency (PyYAML)
• improve readability of configuration reference

current yaml

#########################
# main settings

domain: yourdomain.com
ns1: 127.0.0.1
ns2: 127.0.0.1

#########################
# other configurables

# server address
listen_addr: 0.0.0.0:53

# debug (0 or 1)
debug: 0

# default TTL
ttl: 60

# default latency (in ms)
sleep: 0

# a 3rdparty domain which we don't own
a3rdparty_domain: whatever.com

# known DNS resolvers / clients connecting to us
known_servers:
  127.0.0.1:      localhost
  1.2.3.4:        powerdns1
  1.2.3.5:        powerdns2
  2.3.4.5:        bind-test1
  4.5.6.7:        msdns-test1

#########################

will look like with toml

[main]
domain = 'yourdomain.com'
ns1 = '127.0.0.1'
ns2 = '127.0.0.1'

# server address
listen_addr = '0.0.0.0:53'

# debug (0 or 1)
debug = 0

# default TTL
ttl = 60

# default latency (in ms)
sleep = 0

# a 3rdparty domain which we don't own
a3rdparty_domain = 'whatever.com'

known_servers = """
127.0.0.1   localhost
1.2.3.4     powerdns1
1.2.3.5     powerdns2
2.3.4.5     bind-test1
4.5.6.7     msdns-test1
"""

just a little caveat this will not work to python 3.10 below. so it will be a good idea to update the readme stating this only works with python 3.11 and above.

PR: #3

Hello
Firstly, this is a cool project thank you for making it!
I am trying to run the tests but the md5sum produced by my dig is different than the md5sum that is specified in the test file. For example:
runtest "cgena.1.${domain}"
is supposed to produce the md5sum:
"8bee2a94ebe12cee620a8bfa18169b1d"
but I get something different.
Are you able to run this test in debug mode on your system and share the /tmp/ text file output that produces the target hash? I can then use this to help me troubleshoot why my system is different.
Thank you

Hello
I think I have found an Ethernet padding test bug. For me, the dig command for the 'empty1.yourdomain.com' test returns a Warning banner for queries sent across the internet to a remote instance of PolarDNS, but not for queries to a localhost instance of PolarDNS.

Internet bound query:

$ dig empty1.yourdomain.com @<remote-ip> +tries=1 +timeout=3 -p 53
;; Warning: short (< header size) message received
;; communications error to <remote-ip>#53: timed out

; <<>> DiG 9.18.10 <<>> empty1.yourdomain.com @<remote-ip> +tries=1 +timeout=3 -p 53
;; global options: +cmd
;; no servers could be reached

Local query:

$ dig empty1.yourdomain.com @127.0.0.1 +tries=1 +timeout=3 -p 53
;; communications error to 127.0.0.1#53: timed out

; <<>> DiG 9.18.10 <<>> empty1.yourdomain.com @127.0.0.1 +tries=1 +timeout=3 -p 53
;; global options: +cmd
;; no servers could be reached

This means the md5sum is different so the test might fail. I've checked this difference in Wireshark and I think it is due to extra Ethernet frame padding on the internet query which isn't present on the loopback query. The Warning could be stripped off using the grep -v command in test.sh but the hashes for other 'empty' tests would need to be adjusted to accommodate this.