RE: A wierd RMAN problem. Even wierder than Mladen

  • From: "Mercadante, Thomas F" <thomas.mercadante@xxxxxxxxxxxxxxxxx>
  • To: "'oracle-l@xxxxxxxxxxxxx'" <oracle-l@xxxxxxxxxxxxx>
  • Date: Mon, 8 Mar 2004 08:02:28 -0500

Could these be orphaned net processes that just never went away?  If Rman is
not running, then I would kill'em.
 

Tom Mercadante 
Oracle Certified Professional 

-----Original Message-----
From: Jared.Still@xxxxxxxxxxx [mailto:Jared.Still@xxxxxxxxxxx]
Sent: Sunday, March 07, 2004 11:12 AM
To: oracle-l@xxxxxxxxxxxxx
Subject: Re: A wierd RMAN problem. Even wierder than Mladen



Have you checked support.veritas.com? 

Re Mladens thoughts on 'crosscheck backup': 

The scripts that NetBackup generates are fairly simple. 

I don't believe there's and option to do that when configuring 
a backup with NBU.  Someone could have modified the 
script however. 

Jared 





        "Weaver, Walt" <wweaver@xxxxxxxxxxxx> 
Sent by: oracle-l-bounce@xxxxxxxxxxxxx 


 03/06/2004 12:51 PM 
 Please respond to oracle-l 


        
        To:        <oracle-l@xxxxxxxxxxxxx> 
        cc:         
        Subject:        A wierd RMAN problem. Even wierder than Mladen



Okay, it's not really as wierd as Mladen, but I'm having problems figuring
it out. 
  
We're running 8.1.7.4 on RedHat 7.3, a VALinux 4450 with 2GB RAM, disk is an
EMC, RAID10 exactly like Gaja says it's supposed to be. 
  
It's a backend database server supporting customer websites running our
product. We have five other database servers doing the same thing and they
are fine. 
  
There are approximately 100 customers on the machine. Average load according
to top is 0.50, there is very little load on the machine. 
  
But, recently (the last few days) when RMAN starts its backup process Oracle
exceeds its max# of processes (500) in about a minute. RMAN actually runs
for about ten minutes before this happens but it happens on a regular basis
now. 
  
This morning I kicked RMAN off manually and watched things happening on the
machine. I monitored v$session and kept doing a ps-efll|grep oracle to see
what unix processes were being created. v$session showed just a SYSTEM
process (me) and 4-8 SYS processes (RMAN). But, ps showed a ton of LOCAL=NO
processes, which indicated customers were connecting to the server. But,
when I queried v$session the only connections were me and RMAN. 
  
I killed the RMAN job and everything returned to normal. 
  
So, why would I see so many LOCAL=NO processes but nothing in v$session?
Could it be a listener thing? What could RMAN possibly be doing? 
  
We're using Veritas NetBackup to communicate with RMAN and write everything
directly to tape. The process is identical to how we're doing things on the
other five servers, although they serve fewer customers (8-10, but they're
big customers). 
  
Right now I don't have a clue as to why this is happening. Has anyone else
seen this sort of thing with RMAN? Got any ideas? 
  
Okay, I have to head to the airport now and fly to Idaho. For fun. I'm not
going to worry about this until I get back in a few hours. But, if anyone
has any thoughts on this I'd sure appreciate hearing'em. 
  
Thanks, 
--Walt Weaver 
    Bozeman, Montana 



Other related posts: