Re: DRWR

From: <moizarshad_at_gmail.com>
Date: Thu, 8 Oct 2015 17:06:31 -0500
Message-Id: <A7A05FD5-F87F-4361-829C-E3E2539238EC_at_gmail.com>

I think we went from 2 to 6 dbwr.
What database version are you at?

Thanks,
Moiz

> On Oct 8, 2015, at 4:20 PM, Michael Calisi <oracle455_at_gmail.com> wrote:
> 
> Moiz, thank you.  We did move the redo logs and our SGA is sized correctly.  I feel comfortable that our SGA is sized correctly.   Years back we ran into our pools being over 90% usage and once we clean them up we saw. a major improvement.  I now have jobs run daily that checks the % of our pools.
> 
> Thinking we need additional disk and possible some kernel patches are missing.  I have no intention to switch to ASMM on my production system.    We are using ZSF file system and waiting on Oracle Hardware team before making any changes.  See what they find.
> 
> How many DBWR are you using?    
> 

>> On Thu, Oct 8, 2015 at 5:00 PM, Moizuddin Arshad Mohamed <moizarshad_at_gmail.com> wrote:
>> Hi Michael,
>>
>> I have been in to the similar issue sometime back. Seems to be almost same issue. What file system are you using?
>>
>> We increased the SGA, DBWR and also redo logs were moved to a different disk, faster.
>> We were using Veritas ZFS file systems. After trying all these, still log file sync issue continues......
>>
>> Later Oracle export from ZFS was involved, they applied some kernel level patches (not sure as I am not a sys admin) but prominently, the ZFS file system disk of pools was above 90% usage.... Oracle asked us to get this percentage lower, so system admin added disks, as soon as the percentage was lower than 80%, log file sync issue resolved.
>>
>>
>> Hope this would help.
>>
>> Thanks,
>> Moiz
>>
>>

>>> On Thu, Oct 8, 2015 at 3:41 PM, Michael Calisi <oracle455_at_gmail.com> wrote:
>>> I been dealing with log file sync and write complete wait issues.    
>>> 
>>> Oracle Support is coming back with 
>>> 
>>>  Recommend to set sga_target to non-zero value to enable Automatic shared memory management. This way, you will not run out of free blocks if needed and in turn waits will reduce. 
>>> 
>>> Also, there are 2 db writer processes, I suggest to increase it to 4. This will ensure that the blocks gets copied to datafiles more quickly. This will also help reduce the waits.
>>> 
>>> My understanding is there's no  "golden rule" nor a "golden formula" to increase DBWR. . That you  would configure DBWR parameters only if we really have an issue with DBWR Writes. Even then, you would first look at the (SAN or NAS or DASD) Storage, I/O Channels and OS Parameters and exhaust all other options before looking at these.
>>> We are in the process of moving our redo onto there own drives and checking if there are I/O issues. 
>>> 
>>> I don't feel comfortable changing my production system to ASMM and increasing my dbwr.  before eliminate other areas.
>>> 
>>> What I do want to know is there a way to determine if DBWR need to be increased? will the AWR report provide that information? 
>>> 
> 

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

Received on Fri Oct 09 2015 - 00:06:31 CEST