hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Nick Dimiduk <ndimi...@gmail.com>
Subject Re: Region Server Thread with a Single High Idle CPU
Date Tue, 16 Dec 2014 21:07:55 GMT
1500 reads/sec  to meta with no client activity sounds high, but... how big
is the cluster (how many region servers) and how many thrift gateway
processes are you running? Are you running any other gateways as well (REST
or Thrift2)?

On Tue, Dec 16, 2014 at 11:38 AM, uamadman <uamadman1@gmail.com> wrote:
>
> Currently, hbase-master.com:60010/master-status reports 1500 reads/s when
> no
> clients are connected.
> and 1750-2000 to the .meta when I'm running a load operation. Total R/W
> spikes to 20k/s when adding all the remaining tables together.
>
> I run a multi-core load process that pools 20 cores using happybase to
> interface with the hbase thrift server.
> I typically make the ConPool = 5 + Cores
> happybase.ConnectionPool(size=30,host='') is used to allocate connections.
> results are loaded 10,000 at a time using the table batch functions.
>
> Over all it is really fast.
> I have noticed the hits on the .meta have increased with time. I have yet
> to
> see if they scale with table size.
>
> Thoughts?
> Cheers,
> Jon
>
>
>
>
> --
> View this message in context:
> http://apache-hbase.679495.n3.nabble.com/Region-Server-Thread-with-a-Single-High-Idle-CPU-tp4066876p4066932.html
> Sent from the HBase User mailing list archive at Nabble.com.
>

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