Dnsmasq stops after swap_pager: out of swap space

[Monviech (Cedrik)]

What are those cc interfaces? Looks kinda weird.

What NIC hardware is that? The Chelsio T6225-CR?

Ive never seen that used ever, maybe there is some weird sideffect, especially if netmap is also running on them.

Can you try to deactivate any netmap drivers that attach to these interfaces by disabling zenarmor/intrusion detection and see if dnsmasq still inflates RAM?

[dmurphy]

Some extra data points on my current system

Thank you!  Here's my dnsmasq config; I removed the static reservations since they are repetitive and I didn't feel like masking all of the MAC addresses ... that's just pure laziness on my part.

Nothing crazy in here I don't think!

EDIT: Oh, FYI, the hardware is a DEC2752, vanilla standard ...  the internal interfaces are an 802.3ad active-active LAGG across ax0/ax1;  WAN is on igc0.

Code Select Expand

# DO NOT EDIT THIS FILE -- OPNsense auto-generated file
#

rebind-localhost-ok
stop-dns-rebind

port=53053

# If you want dnsmasq to listen for DHCP and DNS requests only on
# specified interfaces (and the loopback) give the name of the
# interface (eg eth0) here.
# Repeat the line for more than one interface.
interface=lagg0_vlan77,lagg0_vlan99,lagg0,lagg0_vlan91



dhcp-fqdn
domain=home
# This tells dnsmasq that a domain is local and it may answer queries from /etc/hosts
# or DHCP but should never forward queries on that domain to any upstream servers.
local=/home/
local=/dmz/
local=/winhome/
local=/guest/




# On systems which support it, dnsmasq binds the wildcard address,
# even when it is listening on only some interfaces. It then discards
# requests that it shouldn't reply to. This has the advantage of
# working even when interfaces come and go and change address. If you
# want dnsmasq to really bind only the interfaces it is listening on,
# uncomment this option. About the only time you may need this is when
# running another nameserver on the same machine.
bind-interfaces




# Never forward addresses in the non-routed address spaces.
bogus-priv

server=/plex.direct/8.8.8.8
rebind-domain-ok=/plex.direct/

# By  default,  dnsmasq  will  send queries to any of the upstream
# servers it knows about and tries to favour servers to are  known
# to  be  up.  Uncommenting this forces dnsmasq to try each query
# with  each  server  strictly  in  the  order  they   appear   in
# /etc/resolv.conf
strict-order

# Never forward to servers in /etc/resolv.conf
no-resolv






# host entries flushed via dnsmasq_watcher.py [isc] and a dump of the static reservations
addn-hosts=/var/etc/dnsmasq-hosts
addn-hosts=/var/etc/dnsmasq-leases

dns-forward-max=5000
cache-size=10000
local-ttl=1

conf-dir=/usr/local/etc/dnsmasq.conf.d,*.conf

dhcp-range=tag:lagg0_vlan77,192.168.77.50,192.168.77.200,255.255.255.0,86400

domain=dmz,lagg0_vlan77
dhcp-range=tag:lagg0_vlan91,192.168.91.20,192.168.91.50,255.255.255.0,86400

domain=winhome,lagg0_vlan91
dhcp-range=tag:lagg0_vlan99,192.168.99.100,192.168.99.190,255.255.255.0,86400

domain=guest,lagg0_vlan99
dhcp-range=tag:lagg0,192.168.0.100,192.168.0.245,255.255.252.0,86400

domain=home,lagg0
dhcp-range=tag:lagg0,192.168.1.100,192.168.1.250,255.255.252.0,86400

domain=home,lagg0
dhcp-range=tag:lagg0,::,constructor:lagg0,ra-names,ra-stateless,64,86400

domain=home,lagg0
ra-param=lagg0,60,1200

dhcp-range=tag:lagg0_vlan77,::,constructor:lagg0_vlan77,ra-stateless,64,86400

domain=dmz,lagg0_vlan77
ra-param=lagg0_vlan77,60,1200

dhcp-range=tag:lagg0_vlan91,::,constructor:lagg0_vlan91,ra-names,ra-stateless,64,86400

domain=winhome,lagg0_vlan91
ra-param=lagg0_vlan91,60,1200

dhcp-range=tag:lagg0_vlan99,::,constructor:lagg0_vlan99,ra-stateless,64,86400

domain=guest,lagg0_vlan99
ra-param=lagg0_vlan99,60,1200


===== 8< 24 static reservations, all in the same format ... >8 =====
dhcp-host=08:00:20:00:00:00,192.168.0.3,sun-microsystems-forver
===== 8< 24 static reservations, all in the same format ... >8 =====

dhcp-option-force=tag:lagg0_vlan99,3,192.168.99.1
dhcp-option-force=tag:lagg0_vlan99,6,192.168.99.1
dhcp-option-force=tag:lagg0_vlan91,3,192.168.91.1
dhcp-option-force=tag:lagg0_vlan91,6,208.67.222.222,208.67.220.200
dhcp-option-force=tag:lagg0,1,255.255.252.0
dhcp-option-force=tag:lagg0,6,192.168.0.1
dhcp-option-force=tag:lagg0,3,192.168.0.1

# default dns mapped to this server (0.0.0.0)
dhcp-option=6,0.0.0.0


no-ident

Looks like your dnsmasq has a large memory footprint as well - keep an eye on it and see if it continues to grow.

I saw the same post about the dnsmasq 2.92RC3 but didn't have time last night to dig into the patch; seems it's related to parsing some static host files which "shouldn't" be the issue here, but who knows!  Worth checking out.

Thanks so much for digging into this ... thankfully the hourly restart is getting me through but it's certainly not ideal.

[Monviech (Cedrik)]

Nothing weird I can see in the config.

All features I have tested and ran for months without crashes.

Strange.

I could only recommend slowly disabling features until it doesnt happen anymore.

E.g. first disable RA/DHCPv6...check, then DHCPv4... check

That could get to the bottom of this.


------

Also since the known memory leak seems to refer to reading hostnames from files, maybe check if dnsmasq_watcher.py is running?

I refer to this part

# host entries flushed via dnsmasq_watcher.py [isc] and a dump of the static reservations
addn-hosts=/var/etc/dnsmasq-hosts
addn-hosts=/var/etc/dnsmasq-leases