Re: Weirdness with 'ons' process....
- From: "Alex Gorbachev" <gorbyx@xxxxxxxxx>
- To: Mark.Bobak@xxxxxxxxxxxxxxx
- Date: Wed, 1 Nov 2006 00:08:07 -0500
Mark,
I don't know all the functioanlity that's implemented via ONS (aka
Oracle Notification Services) but it's indeed related to FAN events
and all that kitchen. It also involves listeners as they get notified
of node up/down via ONS mechanism. I am not sure but I think that load
balancing info get's propagated via ONS as well and etc.
Anyway, why would it be strange that connections are established from
different source ports? (that's what you mean by 41,446 and ...
63,961?)
On 10/31/06, Bobak, Mark <Mark.Bobak@xxxxxxxxxxxxxxx> wrote:
Hi,
I'm running a 10gR2 (10.2.0.2) 64-bit RAC cluster, three nodes on DL-585s
(configured w/ 4 dual core CPUs and 32GB of RAM). I've been doing some
fairly large-scale load testing, pointing our preproduction environment at
this cluster, and seeing some great performance results.
However, I've been seeing a weird anomaly on two of the three boxes. If I
run:
netstat -an|grep -c ":6101 "
on each of the three nodes, one node reports zero, one reports 41,446 and
one reports 63,961! These connections are all on the local loopback. Where
are these coming from? I know (or I think I know) that this has to do with
'ons'. I haven't found much info on ONS.. I know it has to do with FAN,
which I have not implemented at this point. Is this "normal"? If so, why
the huge difference in distribution of the connections across nodes? (Oh,
also, the status of these thousands of connections is 'TIME_WAIT'.)
As this point, it doesn't seem to be negatively affecting performance, but
it sure seems odd….anyone have any clues for me??
AdvThanksance,
-Mark
--
Mark J. Bobak
Senior Oracle Architect
ProQuest Information & Learning
There is nothing so useless as doing efficiently that which shouldn't be
done at all. –Peter F. Drucker, 1909-2005
--
Best regards,
Alex Gorbachev
The Pythian Group
Sr. Oracle DBA
http://www.pythian.com/blogs/author/alex/
http://blog.oracloid.com
--
//www.freelists.org/webpage/oracle-l
Other related posts:
