I think having MiniAccumuloCluster.start() block until all the services are up is a good thing. I've been seeing somewhat unpredictable behavior also, and this could avoid a lot of questions in the future. The use case of starting the cluster and using a client on it immediately is not typical.
On Mon, Feb 11, 2013 at 5:51 AM, Josh Elser <[email protected]> wrote: > I'd have to agree with you, David. I like MiniAccumuloCluster and think > there is great worth to be had from it. We should do the most we can to > make it easy to use. > > I will say that I imagine it to be difficult to pinpoint exactly when > Accumulo is "up". If there isn't a good way to avoid your exception by > blocking internally, some documentation/warning is a good idea. > > On Sunday, February 10, 2013, David Medinets <[email protected]> > wrote: > > But if people start using the MiniAccumuloCluster without knowing > > their client processing need to wait? What's the harm in letting > > start() do the waiting? Wouldn't that ensure a minimum wait time and > > consistent behavior? > > > > On Sun, Feb 10, 2013 at 11:23 PM, John Vines <[email protected]> wrote: > >> I don't think so, all of our client processes are designed to wait a bit > >> for the server to come up, and this process a similar behavior as our > >> start-all script. > >> > >> Sent from my phone, please pardon the typos and brevity. > >> On Feb 10, 2013 10:59 PM, "David Medinets" <[email protected]> > wrote: > >> > >>> Adding a Thread.sleep(1) after the start() method does seem to > >>> consistently resolve the 'Stream Closed' issue. > >>> > >>> MiniAccumuloCluster la = new MiniAccumuloCluster(tmpDir, > "pass1234"); > >>> la.start(); > >>> Thread.sleep(1); > >>> > >>> Which raises a question. Should the start method wait for the > >>> processes to start before returning? Any reason I shouldn't create a > >>> jira ticket for this? > >>> > >>> On Sat, Feb 9, 2013 at 8:45 PM, Jim Klucar <[email protected]> wrote: > >>> > I wonder if it is just because you're starting and stopping so > quickly. I > >>> > know the Tserver processes take a second to come up, perhaps put a > sleep > >>> in > >>> > for a few seconds and see if that works. > >>> > > >>> > > >>> > On Sat, Feb 9, 2013 at 8:40 PM, David Medinets < > [email protected] > >>> >wrote: > >>> > > >>> >> https://issues.apache.org/jira/browse/ACCUMULO-1058 > >>> >> > >>> >> On Sat, Feb 9, 2013 at 8:04 PM, Christopher <[email protected]> > >>> wrote: > >>> >> > Should probably have documentation and a working example prior to > >>> 1.5.0. > >>> >> Is there already a ticket for it? > >>> >> > >>> > > >
