The stack trace is a typical indicator of trying to run from the source
I wonder if you are setting an environment variable in your install script
which is invalid and then when you rerun manually everything works because
that variable isn't being set.
That error is the monitor logs is an issue scanning the !METADATA table.
Are the tservers online as well? Is the monitor reporting anything at all?
Does it hang when you attempt to reach it or does resolve a blank page,
etc.? Is the monitor process actually running or does it quickly die?
On Sat, May 4, 2013 at 11:19 PM, David Medinets <[EMAIL PROTECTED]>wrote:
> I've been having a problem getting the monitor to run for a little while. I
> figured that I was doing something wrong, but I'm following a shell script
> that used to work. I don't see what the underlying issue is. I'm hoping
> someone might be able to spare some time to help.
> My github project, https://github.com/medined/accumulo_stackscript, has a
> script called install_accumulo_in_home_directory.sh which places hadoop and
> zookeeper in $HOME/software and accumulo in $HOME/bin.
> Something new started happening today which is unrelated to the monitor
> issue. When I run my install script I see
> "java.lang.ClassNotFoundException: org.apache.accumulo.start.Platform" yet,
> right after the script ends, I can manually run the failing command and it
> $HOME/bin/accumulo/bin/accumulo init --clear-instance-name
> --instance-name instance --username root --password secret
> I'm willing to ignore the classpath issue. I'm much more concerned that the
> monitor is not responding.
> I see no warnings or errors in any log file. But I do see this debug
> 2013-05-04 23:06:37,172 [impl.ThriftScanner] DEBUG: Failed to locate
> tablet for table : !0 row : ~err_^@
> Since zookeeper, hadoop and accumulo were all just initialized, I don't
> know why this is happening.
> Any help or suggestions?