hbase-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andrew Purtell <apurt...@apache.org>
Subject Re: ZK rethink?
Date Wed, 08 Apr 2009 18:12:26 GMT

Additional testing does indeed suggest that GC stalls and
insufficiently short default timeout were the causes of my
troubles. This last time I had eyes on the problem and the
JVMs hosting affected HRS went into long GC stalls just prior
to ZK session problems. The correlation is strong. 

I put up an issue for Joey Echeverria's recommendation: 

   - Andy

> From: Ryan Rawson
> Subject: Re: ZK rethink?
> Date: Tuesday, April 7, 2009, 9:49 PM
> Given the problems we've found, perhaps the primary/
> recommended technique for zk should be this?
> Nitay can you do the honors and cross post to zk dev?
> On Apr 7, 2009 9:39 PM, "Nitay"
> It's sounding like folks are interested in pursuing
> Joey's C API with JNI idea. Shall we loop in zookeeper-dev
> to see if they have any opinions on the matter?
> Any other thoughts for solutions we should explore?


View raw message