Posted: Wed Jul 12, 2017 3:19 Post subject: ipv6 connectivity issues: looks like dnsmasq is crashing
Ever since I setup an ipv6 tunnel with Hurricane Electric today, I've noticed intermittent periods of internet connectivity issues across all devices on my network. To try to track this down I was SSH'd to the router (a Linksys 320n converted to e2000) and I was checking the processes with ps. When I have downtime I notice dnsmasq is not in the process list. This condition eventually restarts itself and when the Internet appears to be working, dnsmasq is in the process list again.
Also if I ping by name at the ddwrt root prompt I don't get anything. It just kicks me back to a command prompt.
Do you think this dnsmasq config below is causing the issue? here is my /tmp/dnsmasq.conf file
I checked and syslogd is disabled, so perhaps that's why.
Though with limited memory available I was trying to keep it down to a minimum.
# df
Filesystem 1K-blocks Used Available Use% Mounted on
rootfs 6144 6144 0 100% /
/dev/root 6144 6144 0 100% /
/dev/mtdblock/4 384 196 188 51% /jffs
So I did make a modification to dnsmasq.conf and it seems better now. I have to continue to test.
New config is as follows
interface=br0
domain=ddwrt,192.168.1.0/24,local
enable-ra
ra-param=br0,10,300
dhcp-range=::150,::1EFF,constructor:br0,ra-stateless,ra-names,12h
dhcp-option=option6:dns-server,[::]
dhcp-option=option6:ntp-server,[2001:470:0:50::2]
dhcp-option=option6:domain-search,ddwrt
Now that is set to 12 hour instead of 5 minutes, I'm wondering if that's where the crashes were occurring. I also added ra-stateless as I've seen it in other configs. I'm not sure if thats making a difference or what.
Maybe I'll enable syslogd for a few hours and then disable it, just to see if I log any crashes. I'm just not sure I the space for it.
Well since I made that change to dnsmasq.conf above, I have not logged a single crash, and web browsing did not stall.
I highly recommend looking at papertrailapp.com. Now that I'm sending syslogs there I can alert on anything unusual or just send periodic summaries. I was not aware of all of the IP's from APNIC and RIPE trying to SSH to the router. I would forward a non-standard port, but at work non-standard ports are blocked outbound and then I wouldn't be able to tunnel back home to send WOL and RDP over SSH, or browse the Internet using my connection.