pfBlockerNG not blocking everything in blocklists
-
@gertjan
Yes for some reason every time I run the GRC lookup tool and query my pfSense, pfblockerng is broken afterwards. I can reproduce it every time and when I restart unbound server it is fixed.I am using unbound on pfSense in resolver mode (forwarding option is disabled) and I have no DNS server entries in General Setup tab in pfSense. I have set the option DNS Resolution Behavior to Use local DNS (ignore remote DNS servers).
py_error.log is empty.
dnsbl_error.log shows a few entries[ DNSBL_Abuse_CH - Abuse_ch_hostfile ] Download Fail Firewall and/or IDS (Legacy mode only) are not blocking download. Restoring previously downloaded file [ DNSBL_Abuse_CH - Abuse_ch_hostfile ] Download Fail Firewall and/or IDS (Legacy mode only) are not blocking download. Restoring previously downloaded file
dnsbl_parsed_error.log has some entries like these:
04/6/21 14:16:12,No_Tracking_Hostnames,from,:: from 04/6/21 14:16:12,No_Tracking_Hostnames,hostwinds,0.0.0.0 hostwinds 04/6/21 14:16:12,No_Tracking_Hostnames,hostwinds,:: hostwinds 04/6/21 14:16:30,MS_2,beritanowtk,beritanowtk
-
Hummm.
Everything looks fine to me / identical to me.
Except : I can run that GRC tool as hard as I can, it won't break 'unbound'.
Did you try the old method "unbound mode" ?
-
@gertjan
Good tip @Gertjan. I immediately tested it and low-and-behold. Everything keeps working even if I bombard pfsense with the GRC benchmarking tool. Here the output before and after the GRC testing:> server 192.168.100.1 Default Server: odin.vXXXX.lan Address: 192.168.100.1 > www.conduit.com Server: odin.vXXXXX.lan Address: 192.168.100.1 Name: www.conduit.com Address: 10.10.10.1 > www.vmn.net Server: odin.vXXXXX.lan Address: 192.168.100.1 Name: www.vmn.net Address: 10.10.10.1 > www.ku6.com Server: odin.vXXXXX.lan Address: 192.168.100.1 Name: www.ku6.com Address: 10.10.10.1 >
Looks like there is something the matter with the python integration here. Also as an added bonus in Unbound mode I can see that pfSense is now faster in answering queries compared to the AdGuard home server I have running here. In python mode pfSense was always slower:
Not that the speed difference is that big but still ^^. So in the end something is broken here. I will initiate a reinstall tomorrow unless someone else has another good idea. Thank you very. much for helping Gertjan!
-
Ok I just did a clean install of pfsense and then restored a backup. I have the exact same behavior again. Python mode pfblockerng stops working after I hammer it with GRC DNS benchmark.
For now I am switching over to AdGuard Home. Maybe better luck with pfSense in the next release :(
-
I'm at home now.
Did some testing.
The dns benchmark running, doing it thing.In a console/SSH, I ran this :
tail -f /var/unbound/var/log/pfblockerng/dns_reply.log | grep 'conduit'
According the https://dbl.oisd.nl/ feed ( the oisd.nl feed) :
,www.conduit.com,,0,OISD,DNSBL_Compilation
and also :
,conduit.com,,0,MS_2,DNSBL_BBcan177
I was spamming
www.conduit.com
in a nslookup (Windows 10) session.
Suddenly, during the test :
...... > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0 > www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Nom : www.conduit.com Address: 0.0.0.0
When the 0.0.0.0 came back, I saw this in the 'tail' console :
DNS-reply,Apr 6 23:59:08,servfail,AAAA,AAAA,Unk,www.conduit.com,2001:470:1f13:65c::2001,ServFail,unk DNS-reply,Apr 6 23:59:08,servfail,AAAA,AAAA,Unk,www.conduit.com,2001:470:1f13:65c::2001,ServFail,unk DNS-reply,Apr 6 23:59:08,servfail,AAAA,AAAA,Unk,www.conduit.com,2001:470:1f13:65c::2001,ServFail,unk DNS-reply,Apr 6 23:59:09,servfail,AAAA,AAAA,Unk,www.conduit.com,2001:470:1f13:65c::2001,ServFail,unk DNS-reply,Apr 6 23:59:09,servfail,AAAA,AAAA,Unk,www.conduit.com,2001:470:1f13:65c::2001,ServFail,unk DNS-reply,Apr 6 23:59:09,servfail,AAAA,AAAA,Unk,www.conduit.com,2001:470:1f13:65c::2001,ServFail,unk DNS-reply,Apr 6 23:59:09,servfail,AAAA,AAAA,Unk,www.conduit.com,2001:470:1f13:65c::2001,ServFail,unk DNS-reply,Apr 6 23:59:10,servfail,AAAA,AAAA,Unk,www.conduit.com,2001:470:1f13:65c::2001,ServFail,unk DNS-reply,Apr 6 23:59:10,servfail,AAAA,AAAA,Unk,www.conduit.com,2001:470:1f13:65c::2001,ServFail,unk DNS-reply,Apr 6 23:59:10,servfail,AAAA,AAAA,Unk,www.conduit.com,2001:470:1f13:65c::2001,ServFail,unk DNS-reply,Apr 6 23:59:10,servfail,AAAA,AAAA,Unk,www.conduit.com,2001:470:1f13:65c::2001,ServFail,unk DNS-reply,Apr 6 23:59:11,servfail,AAAA,AAAA,Unk,www.conduit.com,2001:470:1f13:65c::2001,ServFail,unk DNS-reply,Apr 6 23:59:11,servfail,AAAA,AAAA,Unk,www.conduit.com,2001:470:1f13:65c::2001,ServFail,unk DNS-reply,Apr 6 23:59:11,servfail,AAAA,AAAA,Unk,www.conduit.com,2001:470:1f13:65c::2001,ServFail,unk DNS-reply,Apr 6 23:59:11,servfail,AAAA,AAAA,Unk,www.conduit.com,2001:470:1f13:65c::2001,ServFail,unk
and then :
> www.conduit.com. Serveur : priv.my-domain.net Address: 2001:470:xxxx:65c::1 Réponse ne faisant pas autorité : Nom : www.va.conduit.com Address: 198.49.162.233 Aliases: www.conduit.com
DNS-reply,Apr 6 23:59:12,cache,A,A,86399,www.conduit.com,2001:470:xxxx:65c::2001,198.49.162.233,IL
198.49.162.233 popped up.
I guess I can confirm what you saw.
Btw :
I saw :
Assembling DNSBL database...... completed [ 04/7/21 00:05:47 ] TLD: TLD analysis..xxxxxxxxxxx completed [ 04/7/21 00:05:51 ] ** TLD Domain count exceeded. [ 200000 ] All subsequent Domains listed as-is **
during the force update. The OISD feed is BIG : 1,2 million domain names.
I'll test tommorow agagin without the OISD feed, as the base domain name itself "conduit.com" is already in the BBCan 77 feed - as shown above.
edit : without the oisd.nl feed, and thus a rather (for me) small set of blocked domains, env 16962, the "0.0.0.0" blocking of "conduit.com" works.
www.conduit.com is resolved..... -
Interesting! So maybe I am on to something :p.
I got curious and did the test with the BBCan77 feed. The domain conduit.com is indeed there. So I decided to disable the oisd.nl feed and force reload pfblockerNG. After that I bombarded pfSense with the DNS benchmarking tool. Guess what...everything keeps running fine for conduit.com. Multiple reloads of pfBlockerNG, multiple restart of Unbound but everything kept working as expected:
> conduit.com Server: odin.vXXXXX.lan Address: 192.168.100.1 Name: conduit.com Address: 0.0.0.0
Then I enabled the https://dbl.oisd.nl feed did a force reload of pfBlockerNG. I don't have TLD blocking enabled so I don't get that message you have. You can see the log about loading the oisd.nl list:
[ dbl_list ] Downloading update .. 200 OK. ---------------------------------------------------------------------- Orig. Unique # Dups # White # TOP1M Final ---------------------------------------------------------------------- 1251871 1251871 154015 0 0 1097856 ---------------------------------------------------------------------- Saving DNSBL statistics... completed [ 04/7/21 00:31:48 ] ------------------------------------------------------------------------ Assembling DNSBL database...... completed [ 04/7/21 00:31:49 ] Reloading Unbound Resolver (DNSBL python). Stopping Unbound Resolver. Unbound stopped in 2 sec. Additional mounts (DNSBL python): No changes required. Starting Unbound Resolver... completed [ 04/7/21 00:31:53 ] Resolver cache restored DNSBL update [ 1379690 | PASSED ]... completed
And of course now pfBlockerNG is broken and not blocking the domain anymore BUT conduit.com keeps getting blocked by pfblockerng fine:
> server 192.168.100.1 Default Server: odin.vXXXXXX.lan Address: 192.168.100.1 > conduit.com Server: odin.vXXXXXX.lan Address: 192.168.100.1 Name: conduit.com Address: 0.0.0.0 > www.conduit.com Server: odin.vXXXXXX.lan Address: 192.168.100.1 Non-authoritative answer: Name: www.va.conduit.com Address: 198.49.162.233 Aliases: www.conduit.com > www.ku6.com Server: odin.vXXXXXX.lan Address: 192.168.100.1 Non-authoritative answer: Name: 3a1d905b3d90ef4e-cl2.qcloudcjgj.com Address: 118.89.206.86 Aliases: www.ku6.com fa2a686c758fa3d110164df993a0b8c4.qcloudzygj.com > www.vmn.net Server: odin.vXXXXXX.lan Address: 192.168.100.1 Non-authoritative answer: Name: www.vmn.net Address: 54.146.142.3 >
So www.conduit.com is broken but conduit.com is being blocked fine. Even after I run the DNS benchmark tool multiple times conduit.com keeps working.
I am glad you are seeing the same behavior like me. Looks like something is crashing when stressed with multiple (and fast) DNS queries .
-
Same test, back on work.
Home is a 1 Mbytes 'small' VM, work use its own i5, with 8 Mbyyes - although I don't think this matters.First test :
nslookup with full debug :set db2
> www.conduit.com. Serveur : pfsense.my-work-domain.net Address: 2001:470:xxxx:5c0:2::1 ------------ SendRequest(), len 33 HEADER: opcode = QUERY, id = 20, rcode = NOERROR header flags: query, want recursion questions = 1, answers = 0, authority records = 0, additional = 0 QUESTIONS: www.conduit.com, type = A, class = IN ------------ ------------ Got answer (49 bytes): HEADER: opcode = QUERY, id = 20, rcode = NOERROR header flags: response, auth. answer, want recursion, recursion avail. questions = 1, answers = 1, authority records = 0, additional = 0 QUESTIONS: www.conduit.com, type = A, class = IN ANSWERS: -> www.conduit.com type = A, class = IN, dlen = 4 internet address = 0.0.0.0 ttl = 60 (1 min) ------------ ------------ SendRequest(), len 33 HEADER: opcode = QUERY, id = 21, rcode = NOERROR header flags: query, want recursion questions = 1, answers = 0, authority records = 0, additional = 0 QUESTIONS: www.conduit.com, type = AAAA, class = IN ------------ ------------ Got answer (33 bytes): HEADER: opcode = QUERY, id = 21, rcode = SERVFAIL header flags: response, want recursion, recursion avail. questions = 1, answers = 0, authority records = 0, additional = 0 QUESTIONS: www.conduit.com, type = AAAA, class = IN ------------ Nom : www.conduit.com Address: 0.0.0.0
Fine, "www.conduit.com" yields 0.0.0.0
Now I launch a DNS benchmark, and have it finished (30 seconds later)
I "nslookup" again :
> www.conduit.com. Serveur : pfsense.my-work-domain.net Address: 2001:470:xxxx:5c0:2::1 ------------ SendRequest(), len 33 HEADER: opcode = QUERY, id = 22, rcode = NOERROR header flags: query, want recursion questions = 1, answers = 0, authority records = 0, additional = 0 QUESTIONS: www.conduit.com, type = A, class = IN ------------ ------------ Got answer (70 bytes): HEADER: opcode = QUERY, id = 22, rcode = NOERROR header flags: response, want recursion, recursion avail. questions = 1, answers = 2, authority records = 0, additional = 0 QUESTIONS: www.conduit.com, type = A, class = IN ANSWERS: -> www.conduit.com type = CNAME, class = IN, dlen = 9 canonical name = www.va.conduit.com ttl = 86227 (23 hours 57 mins 7 secs) -> www.va.conduit.com type = A, class = IN, dlen = 4 internet address = 198.49.162.233 ttl = 86227 (23 hours 57 mins 7 secs) ------------ Réponse ne faisant pas autorité : ------------ SendRequest(), len 33 HEADER: opcode = QUERY, id = 23, rcode = NOERROR header flags: query, want recursion questions = 1, answers = 0, authority records = 0, additional = 0 QUESTIONS: www.conduit.com, type = AAAA, class = IN ------------ ------------ Got answer (33 bytes): HEADER: opcode = QUERY, id = 23, rcode = SERVFAIL header flags: response, want recursion, recursion avail. questions = 1, answers = 0, authority records = 0, additional = 0 QUESTIONS: www.conduit.com, type = AAAA, class = IN ------------ Nom : www.va.conduit.com Address: 198.49.162.233 Aliases: www.conduit.com
Note the difference :
www.conduit.com resolves to the CNAME "www.va.conduit.com" (domain not listed), and that one resolves to "198.49.162.233" -
Hey Gertjan. This is correct about eventually resolving to a CNAME. I can see it on my end also that after running the GRC benchmark testing tool the domain resolves to a CNAME (www.va.conduit.com) and so pfblockerNG is not blocking it anymore. I think the same thing is happening with "www.ku6.com" and "www.vmn.net".
Can we conclude that the CNAME blocking on pfBlockerNG is not working as expected?
Because the same test on AdguardHome server always gets the domains blocked. I think that is also the reason why I am getting the low percentage in the pfblockerNG dashboard widget (AdguardHome is at 10% blocked DNS requests at the moment).I also think the same thing is happening with other domains as well. Just found out that "diagnostics.meethue.com" gets blocked but after benchmarking with GRC it is not blocked anymore.
I have never seen "diagnostics.meethue.com" in the pfblockerNG DNSBL Block Stats tab being in the list. After running my whole network for 24 hrs on AdGuardhome it is the domain being blocked the most. -
Thanks for the report! Its appreciated!
The issue is that GRC is using a mixture of lowercase and uppercase in their DNS Tests.
I have fixed the code to force everything to lowercase before the domains are validated.
A fix can be downloaded here until the next version is available:
curl -o /var/unbound/pfb_unbound.py "https://gist.githubusercontent.com/BBcan177/1e400f170b7ef7e8cb76b548d6524b31/raw"
Then Restart Unbound
Thanks!
-
Thank you for being so fast in fixing this and the great work you are doing with pfBlockerNG!
With the modifications all is working well :).