What I implied in my first email was that ExecutorType.MASTER_META_SERVER_OPERATIONS should be using its own config parameter. Current code makes it impossible to change the value of "hbase.master.executor.serverops.threads" in .xml without defeating the intention of HBASE-3449.
Cheers On Fri, Feb 11, 2011 at 11:30 AM, Todd Lipcon <[email protected]> wrote: > On Fri, Feb 11, 2011 at 10:59 AM, Ted Yu <[email protected]> wrote: > > > The fix assumed that user wouldn't change the value for > > "hbase.master.executor.serverops.threads" which is not in > hbase-default.xml > > > > Interesting. > > > > > If users are changing advanced undocumented configs, then they deserve what > they get. When you change a config you can only find out about from the > code, you are implicitly accepting a contract that you will follow said > code > and understand the ramifications of your change :) > > -Todd > > > > On Fri, Feb 11, 2011 at 10:03 AM, Todd Lipcon <[email protected]> wrote: > > > > > Nope, that's to get around a bug where HMaster can deadock with itself. > > > Check the svn logs for that section of the code > > > > > > On Fri, Feb 11, 2011 at 10:01 AM, Ted Yu <[email protected]> wrote: > > > > > > > I see: > > > > > > > > > > > > > > > > > > this.executorService.startExecutorService(ExecutorType.MASTER_SERVER_OPERATIONS, > > > > conf.getInt("hbase.master.executor.serverops.threads", 3)); > > > > > > > > > > > > > > > > > > this.executorService.startExecutorService(ExecutorType.MASTER_META_SERVER_OPERATIONS, > > > > conf.getInt("hbase.master.executor.serverops.threads", 5)); > > > > The second might be a typo since default values are different. > > > > > > > > FYI > > > > > > > > > > > > > > > > -- > > > Todd Lipcon > > > Software Engineer, Cloudera > > > > > > > > > -- > Todd Lipcon > Software Engineer, Cloudera >
