Open svetlyak40wt opened 1 week ago
thanks for #73, which adds a continue
restart
I'll keep this issue open though because:
cl-rdkafka::poll-loop
can still exit without being restarted automatically...which leads to this issue againpoll-loop
for a while: librdkafka
allows us to pass in a cffi:defcallback
which will be called from one of the threads it spins up internally...I ended up using the "poll on some fd" approach only because defcallback
was leading to some sbcl segfault/crash at the time
I've added some tracing to the function:
And if there is not broker available, an error is throwed from the the function and it dies:
And attempts to restart consumer don't start a new poll loop:
Probably poll loop should be made a more error proof and run forever?