| Subject: | Smarter control of relay queue via signals or smart submission logic |
<rafl> Apocalypse: relayd wishlist: flush queue on SIGUSR1
<Apocalypse> rafl: Basically, enable/disable no_relay via SIGUSR1?
<@rafl> no, triggering flush of the queue once
<Apocalypse> mmm not that hard to set a flag, check if queue == 0, and
disable relaying once again
<@rafl> i.e. $SIG{USR1} = f { start_relaying(when_queue_empty => f {
stop_relaying }) }
<@rafl> right
<Apocalypse> yuppie
<Apocalypse> let me go file a RT ticket, I'm too busy shaving dzil yaks
to dive into relayd now :)
<@rafl> i don't really care for the signal part
<@rafl> all i want is something to put in my network scripts to flush
stuff shortly after i'm online again
<@rafl> signals just seem the right tool for that
<Apocalypse> yeah, I also thought of an idea to have relayd do
exponential backoff if it detects failed submissions and "cool off"
before trying again
<Apocalypse> i.e. your network goes down, it detects that and retries
every 15m, then once it goes through it resumes "normal" operations
<Apocalypse> However, I haven't got around to doing that because... my
inet is not as flaky as BinGOs' ;)
--
~Apocalypse