hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From stack <st...@duboce.net>
Subject Re: hbase on s3 and safemode
Date Wed, 07 Oct 2009 17:20:47 GMT
Thats interesting to hear.  Keep us posted.

HBase asks the filesystem if its in safe mode and if it is, it parks
itself.  Here is code from master:

    if (this.fs instanceof DistributedFileSystem) {
      // Make sure dfs is not in safe mode
      String message = "Waiting for dfs to exit safe mode...";
      while (((DistributedFileSystem) fs).setSafeMode(
          FSConstants.SafeModeAction.SAFEMODE_GET)) {
        LOG.info(message);
        try {
          Thread.sleep(this.threadWakeFrequency);
        } catch (InterruptedException e) {
          //continue
        }
      }
    }


Then there is hbase's notion of safemode.  It will be in safe mode until it
does initial scan of catalog tables.  The master keeps a flag in zookeeper
while its in safemode so regionservers are aware of the state:

  public boolean inSafeMode() {
    if (safeMode) {
      if(isInitialMetaScanComplete() && regionsInTransition.size() == 0 &&
         tellZooKeeperOutOfSafeMode()) {
        master.connection.unsetRootRegionLocation();
        safeMode = false;
        LOG.info("exiting safe mode");
      } else {
        LOG.info("in safe mode");
      }
    }
    return safeMode;
  }

Have you seen the .META. and -ROOT- deploy to regionservers?  Have you seen
that these regions being scanned in the master log?  (Enable DEBUG if not
already enabled).

Yours,
ST.Ack


On Wed, Oct 7, 2009 at 10:06 AM, Ananth T. Sarathy <
ananth.t.sarathy@gmail.com> wrote:

> We have been running Hbase on a s3 filesystem. It's the hbase regionserver,
> not HDFS since we are using s3.  We haven't felt like it's been too slow,
> though the amount of data we are pushing isn't sufficiently large enough to
> notice yet.
> Ananth T Sarathy
>
>
> On Wed, Oct 7, 2009 at 12:47 PM, stack <stack@duboce.net> wrote:
>
> > HBase or HDFS is in safe mode.  My guess is that its the latter.   Can
> you
> > figure from HDFS logs why it won't leave safe mode?  Usually
> > under-replication or a loss of a large swath of the cluster will flip on
> > the
> > safe-mode switch.
> >
> > Are you trying to run HBASE on an S3 filesystem?  An HBasista tried it in
> > the past and, FYI, found it insufferably slow.  Let us know how it goes
> for
> > you.
> >
> > Thanks,
> > St.Ack
> >
> > On Wed, Oct 7, 2009 at 9:33 AM, Ananth T. Sarathy <
> > ananth.t.sarathy@gmail.com> wrote:
> >
> > > my  regionserver has been stuck in safemode. What can i do to get it
> out
> > > safemode?
> > >
> > > Ananth T Sarathy
> > >
> >
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message