Closed achbed closed 9 years ago
I've currently resorted to a "ping" and service restart if that fails using Drupal. It's checking at most once every 5 minutes. This is why I've added the "pick up where left off playing last time" code to the dev branch - prevents the streams from jumping back to 0 every time a restart happens.
Well, after a server-level update and reboot, everything seems to be working much better (3 days since reboot and no more crash/restarts). If I don't have any more crashes in a few more days, I'll close this issue.
Good news! But ping_server() technic is really nasty although it is the only simple method I found to check if the network is ready (and then waiting silently). Would you see another method for this?
Well, my "ping" checks are actually grabbing the stream status from the icecast server, parsing it, and checking to see if there are any active mountpoints. The symptom is that the source connections close, so icecast reports an empty server status for streams that are sourced from deefuzzer. If this check returns no valid deefuzzer streams, I fire off a root-level service deefuzzer restart
(through a shell script and sudo abstraction layer for security). The restart does a kill -9
before launching another instance of deefuzzer to ensure things are stopped properly.
I'm not sure about network status pings between deefuzzer and the icecast server - ping_server seems to be the least of bad options.
Solved a different way via #45
First of all, I suspect this may be my environment. I'm not sure what's causing it, but DeeFuzzer seems to be crashing about every 13 hours (so far) when playing 10 streams at once. When it dies, there;s nothing reported to the log at all - just silence. I'm unable to see anything in any log that says its either stopping or crashed.
Log snippet during a crash/restart: