[Dnsmasq-discuss] dnsmasq consuming more memory

[Simon Kelley]

The only know memory leak surfaced recently. If you run dnsmasq with the 
--enable-dbus option but without a dbus daemon for it to connect to, it 
will try endless to connect, leaking memory each time.

It costs very little to restart dnsmasq, so a workaround is a cron job 
to do that. 2.80 is pretty old, I'd like you to test a more recent 
version before putting effort into chasing this.


Cheers,

Simon.

On 11/03/2022 10:20, shashikumar Shashi wrote:
> Hi Simon Kelley,
> 
> dnsmasq is running continuously for around five months, we had 
> around 150 clients. After 5 months dnsmasq got killed by the kernel, due 
> to cgroup memory limit.
> It utilizes around 2GB of memory.
> Are there any known issues, we are using 2.80 dnsmasq version?
> 
> kernel: [ pid ]   uid  tgid total_vm      rss nr_ptes nr_pmds swapents 
> oom_score_adj name
> kernel: [16698]     0 16698     6350     3088      15       3        0   
>            0 dnsmasq
> kernel: [16717]     0 16717     6350     1219      10       3        0   
>            0 dnsmasq
> kernel: [16751]     0 16751     1326      947       5       3        0   
>            0 python
> kernel: Memory cgroup out of memory: Kill process 16698 (dnsmasq) score 
> 5 or sacrifice child
> kernel: Killed process 16717 (dnsmasq) total-vm:25400kB, 
> anon-rss:4876kB, file-rss:0kB, shmem-rss
> kernel: oom_reaper: reaped process 16717 (dnsmasq), now anon-rss:0kB, 
> file-rss:0kB, shmem-rss:0kB
> systemd[1]: dnsmasq_vrf at swns.service: Unit entered failed state.
> systemd[1]: dnsmasq_vrf at swns.service: Failed with result 'signal'.
> 
> 
> Thanks,
> Shashi
> 
> _______________________________________________
> Dnsmasq-discuss mailing list
> Dnsmasq-discuss at lists.thekelleys.org.uk
> https://lists.thekelleys.org.uk/cgi-bin/mailman/listinfo/dnsmasq-discuss