[darkice] Re: watch darkice errors and restart if necessarily

  • From: Hungerburg <pch13@xxxxxxxxx>
  • To: darkice@xxxxxxxxxxxxx
  • Date: Wed, 18 Jan 2012 00:48:53 +0100

Am 2012-01-17 20:48, schrieb JoergSorge:

we need only to running multiple streams from ONE source.

If there is only one source, you do not need two instances of darkice, to stream to two servers. darkice can be configured for that.

Looking at your script though: you only install one watchdog, for the "external" connection, that is. So the "internal" connection is stable? And having two instances make sense, after all.

Which points at the one thing this thread is all about: darkice does not handle flaky networking gracefully.

As far as I remember: When the uplink gets congested, some sink cannot consume all available data and a barely noticeable dropout occurs on the other end, yet this data will continue to exist in the ring buffer and mess with its pointer and structure, probably causing it to shrink. In my experience, after several of those dropouts, all of a sudden the stale data will spill out of the ring buffer, pointer and structure being reset, and it all will start from the beginning.

--
peter


Other related posts: