DNS Resolver saturated bandwidth causing no internet

justsomeguy6575 · Aug 9, 2016, 2:22 AM

I'm not sure what the issue is but it seems to be related to using DNS Resolver instead of Forwarder.
Going back to 2.2.6 didn't help anything. Again switching to DNS Forwarder seems to fix it.
I connected my pc directly to my cable modem temporarily, bypassing pfsense. Saturated my download bandwidth and was still able to load 5 webpages simultaneously.

Doing a continuous ping to google gives me average ~30ms response time, with bandwidth saturated that jumps to 1500-2000ms but still responds without any timeouts. With that much delay is it possible the resolver gives up before getting a response?

RonpfS · Aug 9, 2016, 3:22 AM

How did you configure the Resolver ? Did you disable DHCP Registration and Static DHCP ?
Is there a custom Options line```
server:include: /var/unbound/pfb_dnsbl.conf

Do you see anything in the Resolver Log? You may have to restart unbound in order for it to log.

Before restoring the config to a new install you should edit it and disable pfBlockerNG, DNSBL
For unbound to start without DNSBL modification change the config unbound section

<unbound><active_interface>lan,lo0</active_interface>
<outgoing_interface>wan</outgoing_interface>
<custom_options>c2VydmVyOmluY2x1ZGU6IC92YXIvdW5ib3VuZC9wZmJfZG5zYmwuY29uZg==</custom_options></unbound>

to


or in a shell run```
touch /var/unbound/pfb_dnsbl.conf
```.

When you installed on 2.2.6 did it install pfblockerNG ? If /var is running low, the installation might fail when downloading the MaxMind database.

justsomeguy6575 · Aug 9, 2016, 3:59 AM

thanks for replying.

I didn't specify in the last post but when going back to 2.2.6 it was a totally stock install, no pfblocker or any other other packages not installed by the firewall itself and no custom options in the resolver. I had also done a complete fresh install of 2.3.2 without any packages installed. In both cases the resolver was configured just how it comes by default. I have also had dhcp reg and static reg enabled and disabled. Should it be one way or the other?

I tested it again recreating the problem and watching the log and like you mentioned nothing showed up during that time. I then went back to the resolver and restarted the service. This is what showed up in the log. Not sure what it all means but nothing really looks out of place.

Aug 8 20:41:14 	unbound 	25924:0 	info: start of service (unbound 1.5.9).
Aug 8 20:41:14 	unbound 	25924:0 	notice: init module 1: iterator
Aug 8 20:41:14 	unbound 	25924:0 	notice: init module 0: validator
Aug 8 20:41:13 	unbound 	32784:0 	info: 64.000000 128.000000 6
Aug 8 20:41:13 	unbound 	32784:0 	info: 32.000000 64.000000 63
Aug 8 20:41:13 	unbound 	32784:0 	info: 16.000000 32.000000 132
Aug 8 20:41:13 	unbound 	32784:0 	info: 8.000000 16.000000 122
Aug 8 20:41:13 	unbound 	32784:0 	info: 4.000000 8.000000 79
Aug 8 20:41:13 	unbound 	32784:0 	info: 2.000000 4.000000 55
Aug 8 20:41:13 	unbound 	32784:0 	info: 1.000000 2.000000 51
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.524288 1.000000 44
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.262144 0.524288 90
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.131072 0.262144 166
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.065536 0.131072 166
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.032768 0.065536 99
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.016384 0.032768 94
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.008192 0.016384 2
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.004096 0.008192 3
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.002048 0.004096 1
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.000256 0.000512 1
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.000032 0.000064 1
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.000000 0.000001 220
Aug 8 20:41:13 	unbound 	32784:0 	info: lower(secs) upper(secs) recursions
Aug 8 20:41:13 	unbound 	32784:0 	info: [25%]=0.041622 median[50%]=0.218322 [75%]=6.6962
Aug 8 20:41:13 	unbound 	32784:0 	info: histogram of recursion processing times
Aug 8 20:41:13 	unbound 	32784:0 	info: average recursion processing time 6.040819 sec
Aug 8 20:41:13 	unbound 	32784:0 	info: server stats for thread 1: requestlist max 58 avg 11.2079 exceeded 0 jostled 0
Aug 8 20:41:13 	unbound 	32784:0 	info: server stats for thread 1: 2333 queries, 938 answers from cache, 1395 recursions, 0 prefetch
Aug 8 20:41:13 	unbound 	32784:0 	info: 32.000000 64.000000 13
Aug 8 20:41:13 	unbound 	32784:0 	info: 16.000000 32.000000 24
Aug 8 20:41:13 	unbound 	32784:0 	info: 8.000000 16.000000 26
Aug 8 20:41:13 	unbound 	32784:0 	info: 4.000000 8.000000 20
Aug 8 20:41:13 	unbound 	32784:0 	info: 2.000000 4.000000 9
Aug 8 20:41:13 	unbound 	32784:0 	info: 1.000000 2.000000 9
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.524288 1.000000 12
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.262144 0.524288 46
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.131072 0.262144 72
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.065536 0.131072 53
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.032768 0.065536 20
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.016384 0.032768 25
Aug 8 20:41:13 	unbound 	32784:0 	info: 0.000000 0.000001 15
Aug 8 20:41:13 	unbound 	32784:0 	info: lower(secs) upper(secs) recursions
Aug 8 20:41:13 	unbound 	32784:0 	info: [25%]=0.0976857 median[50%]=0.238478 [75%]=3.33333
Aug 8 20:41:13 	unbound 	32784:0 	info: histogram of recursion processing times
Aug 8 20:41:13 	unbound 	32784:0 	info: average recursion processing time 4.845544 sec
Aug 8 20:41:13 	unbound 	32784:0 	info: server stats for thread 0: requestlist max 44 avg 4.40407 exceeded 0 jostled 0
Aug 8 20:41:13 	unbound 	32784:0 	info: server stats for thread 0: 442 queries, 98 answers from cache, 344 recursions, 0 prefetch
Aug 8 20:41:13 	unbound 	32784:0 	info: service stopped (unbound 1.5.9).
Aug 8 19:07:37 	unbound 	32784:0 	info: start of service (unbound 1.5.9).

RonpfS · Aug 9, 2016, 4:07 AM

From this point, unbound should log when it restart (DNSBL update will restart unbound)

About the registration :
@BBcan177:

Some recommendations:

The DNS Resolver can also be used in 'Forwardering mode'; however its best to not use this 'Forwarding mode' and keep it in 'resolver mode' as this will query the Root DNS servers for the DNS queries instead of relying on an ISPs DNS etc…

If you use the 'DNS Resolver Forwarder mode', only configure 'DNSSEC' if the configured DNS servers support DNSSEC. The enabling of 'DNSSEC' to harden your DNS security is highly recommended.

Disable the two "DHCP registrations" checkboxes, unless you really require those options.

RonpfS · Aug 9, 2016, 4:28 AM

So what is on the system now ? no pfblockerng installation? or did you install and removed it ?
What does Diagnostic / System activity shows? Is the system busy?
How much RAM, disk space? what kind of CPU?
Anything weird in the Firewall Logs ?

justsomeguy6575 · Aug 9, 2016, 4:58 AM

At the moment I've restored my backup image of my original 2.3.2 install with pfblockerng and all my nat/rules since even a completely raw install without any of that made no difference. When testing with 2.2.6 and clean 2.3.2, no rules or packages were added. It was just install pfsense from flash drive set lan/wan interfaces and then test the problem.

hardware is:
Core2Duo 6420
4GB RAM
160GB HD

Here is what System Activity shows while saturating my bandwidth:

last pid: 31177;  load averages:  0.09,  0.04,  0.02  up 0+03:18:47    21:43:55
158 processes: 3 running, 122 sleeping, 33 waiting

Mem: 42M Active, 99M Inact, 196M Wired, 283M Buf, 3588M Free
Swap: 8192M Total, 8192M Free

  PID USERNAME PRI NICE   SIZE    RES STATE   C   TIME    WCPU COMMAND
   11 root     155 ki31     0K    32K RUN     1 190:55  94.97% [idle{idle: cpu1}]
   11 root     155 ki31     0K    32K RUN     0 191:39  94.29% [idle{idle: cpu0}]
   12 root     -92    -     0K   528K WAIT    0   2:10   9.18% [intr{irq21: skc0 uhci}]
   12 root     -92    -     0K   528K WAIT    1   1:45   8.25% [intr{irq16: skc1 uhci}]
 7503 root      21    0   262M 31908K piperd  0   0:00   0.39% php-fpm: pool nginx (php-fpm)
   12 root     -60    -     0K   528K WAIT    0   0:43   0.10% [intr{swi4: clock}]
    0 root     -16    -     0K   192K swapin  0   0:37   0.00% [kernel{swapper}]
    5 root     -16    -     0K    16K pftm    0   0:05   0.00% [pf purge]
74270 root      20    0   224M 33304K nanslp  0   0:03   0.00% /usr/local/bin/php -f /usr/local/pkg/pfblo
74759 root      20    0   224M 33324K nanslp  1   0:03   0.00% /usr/local/bin/php -f /usr/local/pkg/pfblo
25924 unbound   20    0 51036K 25616K kqread  0   0:01   0.00% /usr/local/sbin/unbound -c /var/unbound/un
   15 root     -16    -     0K    16K -       1   0:01   0.00% [rand_harvestq]
43973 root      52   20 17000K  2560K wait    0   0:01   0.00% /bin/sh /var/db/rrd/updaterrd.sh
85333 root      20    0 14516K  2316K select  1   0:01   0.00% /usr/sbin/syslogd -s -c -c -l /var/dhcpd/v
28401 root      20    0 39136K  7204K kqread  1   0:01   0.00% nginx: worker process (nginx)
 8130 root      20    0 19108K  2376K nanslp  1   0:01   0.00% [dpinger{dpinger}]
  271 root      22    0   262M 24928K kqread  0   0:01   0.00% php-fpm: master process (/usr/local/lib/ph
28632 root      20    0 39136K  7164K kqread  0   0:01   0.00% nginx: worker process (nginx)

RonpfS · Aug 9, 2016, 5:13 AM

You have plenty of free memory

From the look of it, the system looks idle except for these 2 processes
9.18% [intr{irq21: skc0 uhci}]
8.25% [intr{irq16: skc1 uhci}]

% looks like very high too me, maybe the slowdown is related to interrupt processing of you NIC?

So this is with DNSBL running?

In Diagnostics / Command Prompt execute

ps -axwwwll | grep pfb

this is was I get on my system

   0 18599     1   0  20  0   12856   4224 kqread   S     -      0:12.46 /usr/local/sbin/lighttpd_pfb -f /var/unbound/pfb_dnsbl_lighty.conf
   0 53399     1   0  20  0   38376  10324 nanslp   S     -      6:07.15 /usr/local/bin/php -f /usr/local/pkg/pfblockerng/pfblockerng.inc dnsbl
   0 99209 79568   0  22  0   10460   2084 wait     S     -      0:00.00 sh -c ps -axwwwll | grep pfb 2>&1
   0 99758 99209   0  22  0   10264   1868 piperd   S     -      0:00.00 grep pfb

justsomeguy6575 · Aug 9, 2016, 5:29 AM

Yes that was with DNSBL running.

Here's the result of 'ps -axwwwll | grep pfb'

   0  1768     1   0  20  0  40260  6164 kqread   S     -    0:00.20 /usr/local/sbin/lighttpd_pfb -f /var/unbound/pfb_dnsbl_lighty.conf
   0 60590 98301   0  21  0  17000  2508 wait     S     -    0:00.00 sh -c ps -axwwwll | grep pfb 2>&1
   0 60988 60590   0  21  0  18740  2244 piperd   S     -    0:00.00 grep pfb
   0 74270     1   0  20  0 229352 33308 nanslp   S     -    0:03.94 /usr/local/bin/php -f /usr/local/pkg/pfblockerng/pfblockerng.inc dnsbl
   0 74759     1   0  20  0 229352 33324 nanslp   S     -    0:03.91 /usr/local/bin/php -f /usr/local/pkg/pfblockerng/pfblockerng.inc dnsbl

I may have a couple old linksys or netgear cards laying around that I'll try putting in the system tomorrow and see how it responds. Right now they are gigabit D-link cards.

RonpfS · Aug 9, 2016, 5:40 AM

Well it's weird that you have 2 /usr/local/pkg/pfblockerng/pfblockerng.inc dnsbl running
Try disabling DNSBL, ps should have no pfblockerng at all

justsomeguy6575 · Aug 14, 2016, 12:19 AM

I've kind of given up on making this work.
I've swapped network cards twice with old linksys cards and with old 3com cards. Done complete clean install of pfSense with no added packages. The results are still the same. I have to enable forwarding otherwise DNS queries just don't work if my internet bandwidth is near saturated. However when saturated if forwarding is enabled dns works and pages will load, again slower but they still work. Hopefully the next release will have some improvement.