hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michael Segel <michael_se...@hotmail.com>
Subject RE: Tuning question...
Date Thu, 08 Apr 2010 19:21:01 GMT

Well I went back to u17 from u18.

I'm still having issues where I have regions servers going 'zombie' That is, the regionserver
starts a new session but the jobs are trying to attach to the old session and are getting
a connection refused. (The old session is not up to allow connections.)

I'm playing with the newSize and maxNewSize directives as I run some jobs and I'm also trying
to see what my developers are doing. (Lots of moving parts so its hard to find the error.)
Already went through the 17 tuning tips and the GC data. 

Note: I moved the ZK off my data nodes. While I didn't think that they were an issue, moving
them removed one moving piece. The ZKs don't seem to be eating too many resources on their

I'm trying to see what could be causing my issues. Looks like GC is the main culprit, but
I need to be sure.

Anyone know issues w u17? Or if u19 is safe?



> Date: Wed, 7 Apr 2010 21:13:26 -0700
> Subject: Re: Tuning question...
> From: ryanobjc@gmail.com
> To: hbase-user@hadoop.apache.org
> I'm using u14.
> Old school!
> Considering how buggy u18 was, I'm not super inclined to upgrade.  u14
> runs fine w/o crashing.
> On Wed, Apr 7, 2010 at 9:06 PM, Stack <stack@duboce.net> wrote:
> > On Wed, Apr 7, 2010 at 3:36 PM, Michael Dalton <mwdalton@gmail.com> wrote:
> >> Correct me if I'm wrong on any of this, but is
> >> escape analysis safe at all to turn on in production given that it's totally
> >> disabled now in the most recent Java builds?
> >>
> > Thats sounds right.  Ryan, you are running u16?  Its off in u19 too I believe.
> > St.Ack
> >
The New Busy is not the too busy. Combine all your e-mail accounts with Hotmail.
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message