Closed fractaloop closed 10 years ago
"Child termination detected" is supposed to print some more info. Does it say anything ? Is there any exception / stacktrace ?
I don't get much. What packages should I turn on debug logging for?
2014-04-29 17:08:18.077 ERROR com.redis.RedisConnection akka.tcp://Paparazzi@paparazzi0-int:2553/user/redis-client-8 - Child termination detected: Actor[akka://Paparazzi/user/redis-client-8/pipeline#-694129782]. Reconnecting in 100 ms...
I was thinking of https://github.com/debasishg/scala-redis-nb/blob/master/src/main/scala/com/redis/RedisConnection.scala#L118 .. which has this x
that I thought may give some information. But I was mistaken. Looks like we need to dig into the IO layer to see what's happening.
I did some deep debugging and its definitely a Terminated(
connection)
in TcpPipelineHandler
. I'll keep looking.
This one is really strange. I added logging to TcpPipelineHandler, and it does in fact reconnect, and the client tells me it connected. After this, no further commands to Redis work. I haven't been able to reliably reproduce it.
I'm going to close this for now. I think this was an error in my own app and unrelated to reconnections. Sorry for the noise.
Fractaloop, I may be running into the same issue as you described. Can you tell me what the issue and maybe fix was on your application side?
@bearrito I think it was because my Lua script had an error. :sweat:
So I've deployed the most recent 0.5 release with the reconnections, but the underlying issue of the client termination still plagues me. Worse, further calls to
IO(Tcp) ! Connect(remote)
fail and the JVM must be restarted for future connections.This seems like it might be a bug in the IO layer, or perhaps the TcpPipeline, but the client dies so anonymously ("Child termination detected") it masks the underlying cause.
I am using a good deal of EVALSHA which might be causing pipeline problems, but the redis log doesn't have any errors. Do y'all have any idea what might be causing this?