<karlp>
stintel: I actually _like_ the predictable interface names, even if they're longer. I find them vastly clearer to distinguish than "eth0,1,2"
* opty
got so used to unpredictable ones :)
robimarko has joined #openwrt-devel
csrf has quit [Quit: Leaving]
<karlp>
bleh. g_ether gadget happily works with windows. g_cdc, which is "g_ether plus an acm console" actually leaves out the bits of g_ether that works with windows.
<aiyion>
svanheule: anything left on my side regarding the GPIO fix?
<will[m]>
is there a reason why dnsmasq runs as root in openwrt? by default it runs as "nobody"
<robimarko>
Cause, OpenWrt runs as root only
<olmari>
will: Unless something is changed recent days, openwrt has generally been root-only anyway
<will[m]>
In principle yeah but on my stripped down openwrt it runs as nobody and I wonder if that's causing problems, and where that user selection is performed to see if maybe there's a hint as to why it's important
<will[m]>
I'm not using the normal init stuff or ubus in my minimal environment just running it directly
<olmari>
Do you see any problems?
<slh>
dnsmasq itself is running as user 'dnsmasq' for the last couple of releases already
<will[m]>
It's been simply failing to respond to requests after a few days of uptime, and won't recover upon service restart, but weirdly does recover when I bounce eth0
<will[m]>
Oh interesting I'm a few releases behind, haven't prioritized the bleeding edge lately
<slh>
don't remember if that was introduced with 18.06 or 19.07, but it's been a while - and it's also running under ujail as non-privileged user these days
<will[m]>
It could also be that since I squash ubus, dnsmasq gets upset that it can't talk to ubus and some resource locks up after awhile. I'm trying a version without ubus now
<olmari>
will: is there some special reason you strip owrt this down?
<slh>
try a current release first, for security reasons alone
<olmari>
I mean, whole openwrt is stripped down already as an concept ina in practive too very tiny ;)
<will[m]>
I'm running a custom binary that coordinates network services so it's useful to have a stripped router OS and package ecosystem but it would conflict with ubus and uci and stuff if I let the OS just do its thing naturally
cbeznea has quit [Quit: Leaving.]
cbeznea has joined #openwrt-devel
<will[m]>
Thanks for your feedback!
csrf has joined #openwrt-devel
<olmari>
I'm not saying you're inherently wrong in there... Just that quite highly special and magicery to strip it down by core parts
<slh>
and it's only a sensible approach, if you can still keep up with -stable and its security updates in time
danitool has joined #openwrt-devel
Acinonyx has joined #openwrt-devel
<will[m]>
Oh yeah absolutely. Just checking in, thanks!
<will[m]>
Sometimes there's a very specific reason and it's well known, versus "shrug"