Re: more info - Re: 100% CPU on Startup and Shutdown

opus wrote:
> Pretty much confirmed that this happens during redo log swaps. Also, there's
> another database on the same server that does not have any issues.
>
> Thx,
> D.
>
> opus <opus_at_stopspamvalleyg.net> wrote:
> >Hello...
> >
> >Oracle 8.1.7 on Win2k-Serv SP2 P3-1.2 1GB RAM. Lots of freespace on all
> >partitions.
> >
> >This database has been working flawlessly for several years. Recently I had to
> >turn off the computer to swap the UPS. Upon restarted the server the Oracle
> >instance now goes to 100% CPU usage on instance startups and shutdowns,
> >dramatically slowing down the process. Instance startup takes upwards of 10
> >minutes. This happens whether using DBA Studio or oradim. I suspect there may
> >be slowing down during redo log switches, too.
> >
> >No other changes to the database that I'm aware of. The server has been
> >rebooted several times in the past without incident.
> >
> >I'm not very experienced as a DBA, so please forgive me if I ask a silly
> >question or two.
> >
> >Windows event logs and oracle logs are clean, though I admit I may not be
> >looking everywhere or reading them correctly. I'm guessing the first thing
> >might be get new memory.
> >
> >It's a mission critical business database, so I'm very concerned. Any
> >suggestions?
> >
> >Thx,
> >D.
> >

First: PLEASE do NOT top post.

Second: You do not mention your disk configuration, and you only just mentioned a second instance on this same server. Why do you insist upon running two instances on a single Windows 2000 server? How large are the SGAs? Are the same disks used for both instances, and in the same capacity for each (all redo logs for both instances on the same disk, datafiles for both on the same disks, etc.)? What else may have changed on this server? Is it configured for automatic updates from Microsoft? If so, how may times has this server been restarted due to installed updates? What patch level is the database software, as you do not mention any patches having been applied?

Certainly this COULD be a memory problem, however it's probably not with the quality of the memory but with the quantity. You state this second instance has no performance problems; could it be this other instance starts before your 'mission critical' instance, and thus consumes memory once solely allocated to your 'mission critical database'?

You have left many questions unanswered, and as a result no answers can be forthcoming. Provide the requested information if you truly wish a useful response.

David Fitzjarrell Received on Mon Jun 06 2005 - 14:55:33 CDT