Steve, I meant hadoop-env.sh.
Also, I have a thread called 'Object Heap Error' on Mar 11 03:42 in the Accumulo User Mailing List Archive. That thread addresses the Accumulo/Zookeeper issue. You might want to review that as well. Looks like I had to move the zookeeper jar into the accumulo/lib to get it working. Don Resnik On Tue, Jun 12, 2012 at 6:48 PM, Steven Troxell <[email protected]>wrote: > Hi all, > > If anyone has experience with Accumulo on Ec2, and can help me out, I'd > greatly appreciate it. > > I have a client application that requires only a zookeeper server config > entry to connect to accumulo. Set to "localhost" on my personal box this > works fine. It seems fairly well-documented ( > http://search-hadoop.com/m/sPdqNFAwyg2 for example) that zookeeper and > EC2 do not play nice. I've tried some of the proposed solutions such as > using internal Ec2 hostnames in the zookeeper server config entry all to > still get the connection refused issues. > > > Moving my client into EC2 does not seem to be a viable option, because I > have so far been unsuccessful utilizing the ruby net/https to interact with > it. I get obscure SSL errors when trying to connect from EC2 to > localhost, and Connection reset by peer when trying to connect from > outside EC2. > > My question is, does anyone have a working example of how to specify > Zookeeper instances/servers for Ec2? Alternativley, would it make more > sense to move accumulo off EC2, and point Accumulo towards it instead? Or > would this likely run into the same problem, cause at some point, zookeeper > would then need to connect in to EC2 to accumulo? > > Thanks in advance for any/all help. > > > -- Don Resnik Expert Consultants Inc. 410-854-1756 (desk) 240-381-9412 (cell) 301-957-2270 (ECI office)
