> On 15 Jan 2018, at 23:27, Simon Kelley <si...@thekelleys.org.uk> wrote:
>> Beyond “gaaahhhhhh why didn’t I think of SIGINT”….. excellent.  Understand 
>> the reasoning, agree, running chez Kevin and backport for LEDE master 
>> submitted.
> and there's still SIGQUIT available!
> Out of interest, how does the LEDE plumbing deal with a restart of
> dnsmasq _after_ ntp  has established lock?

There’s an ntp hotplug script that creates a ‘time is valid’ flag file 
(/var/state/dnsmasqsec - being in /var means actually in tmp hence ram).  If 
the file doesn’t exist already on stratum change then a) it gets created and b) 
SIGINTs dnsmasq.   dnsmasq startup changes too… if the file doesn’t exist then 
it gets started with ‘—no-dnssec-timestamp’ expecting to be SIGINT’d by the 
hotplug script.  If dnsmasq gets (re)-started and ntpd hotplug has created our 
‘time valid’ file, then dnsmasq is started *without* —no-dnssec-timestamp’.  
There’s a whole raft of logic related to whether or not we’re using dnssec and 
quite what ntp client is being used.

The SIGINT support was committed to LEDE/openwrt master (rather than CC, 1701 
or whatever) around 60 minutes ago.


Kevin D-B

012C ACB2 28C6 C53E 9775  9123 B3A2 389B 9DE2 334A

Attachment: signature.asc
Description: Message signed with OpenPGP

Dnsmasq-discuss mailing list

Reply via email to