Re: received more blocks than served?

  • From: Dan Norris <dannorris@xxxxxxxxxxxxx>
  • To: 조동욱 <ukja.dion@xxxxxxxxx>
  • Date: Tue, 08 Jul 2008 11:21:32 -0600

I think Mark's suggestion about the different start times is probably the culprit. Flawed logic on my part--thanks Mark for pointing it out. Guess I was having a Monday :-|

Dan

조동욱 wrote:
Can you compare the result
- with the total summation of "received" and "served" blocks, including cr blocks?
- and with the summation of cr blocks only?

Dion Cho



2008/7/8 <dannorris@xxxxxxxxxxxxx>:
I have a 3-node 10.2.0.3 RAC cluster on RH4. I was looking for
verification that I hadn't dropped any blocks between instances during GC
operations. So, I wanted to get a sum of all blocks sent, a sum of all
blocks received and they should be very close to equal.

What I see below has me perplexed. How can there be more blocks received
than served? That is, how can an instance receive more information than
was sent? Can a single block be served to multiple destinations? I don't
think so, but maybe that's where the magical block synthesis is coming
from?

It's a 3-node cluster, so there's no easy 1-to-1 mapping to be done.

I'm open for ideas...

 1  select sum(b2.value) "GCS CURR BLOCKS RECEIVED",
 2  sum(b1.value) "GCS CURR BLOCKS SERVED",
 3  (sum(b1.value)-sum(b2.value)) "SERVED - RECEIVED"
 4  from gv$sysstat b1, gv$sysstat b2
 5  where b1.name = 'gc current blocks served'
 6  and   b2.name = 'gc current blocks received'
 7* and   b1.inst_id = b2.inst_id

GCS CURR BLOCKS RECEIVED GCS CURR BLOCKS SERVED    SERVED - RECEIVED
------------------------ ---------------------- --------------------
              628072519              622037942             -6034577

Dan



--
//www.freelists.org/webpage/oracle-l





--
=====================================
EXEM. The Performance Artist Group
DB기술본부/ 교육컨텐츠팀 / 수석 조 동욱
서울시 강남구 역삼동 아주빌딩 902호
Tel: 02-6203-6300
ukja@xxxxxxxxx, ukja.dion@xxxxxxxxx
Blog: http://ukja.tistory.com
Wiki: http://wiki.ex-em.com
http://wiki.ex-em.com/index.php/Oracle_is_mad
=====================================
-- //www.freelists.org/webpage/oracle-l

Other related posts: