Re: EMC Storage Array Issue

It seems silly to get into a semantic argument over this. My point was that Oracle in fact does use asynchronous io as far as the POSIX definition of aio. Just because the ocntrolling thread looks in on the status of the aiocb when it is signaled upon completion doesn't make it any less asynchronous.

On Monday, April 15, 2002, at 06:23 PM, Anjo Kolk wrote:

> George,
>
> Those writes are unbuffered (so synchronous) and each thread will write
> them
> out and will wait for the write to finish and then update the AIO
> control
> block in the parent thread to indicate that the I/O took place. The
> parent
> thread will check this control block to make sure that the I/O really
> happened. If there is an error or some time out experid and the i/o
> didn't
> complete, the parent thread will assume an error.
>
> Anjo.
>
>
> George Schlossnagle wrote:
>
>> On Monday, April 15, 2002, at 02:08 PM, Anjo Kolk wrote:
>>
>>> Claudio,
>>>
>>> All writes in Oracle are synchronous, they are just fired
>>> asynchronously. So
>>> even the DBWR writes synchronously.
>>
>> My dbw uses asynchronous writes.... I'm writing to qio files, which
>> means that they bypass buffer cache, but I think it's misleading to
>> call
>> those 'synchronous writes', they're asynchronous, unbuffered writes.
>>
>> 14:28:09(george_at_core-0)[~]> truss -tkaio -twrite -tread -p `ps -ef |
>> grep dbw | grep -v grep | awk '{print $2}'`
>> kaio(AIOWRITE, 396, 0xB0258000, 8192, 0x4FBF6000, 0x01974EC8) = 0
>> kaio(AIOWRITE, 308, 0xB6BFE000, 8192, 0x48478000, 0x01B20518) = 0
>> kaio(AIOWRITE, 396, 0xB921A000, 8192, 0x249E8000, 0x019C5558) = 0
>> kaio(AIOWRITE, 396, 0xB4584000, 8192, 0x24F7E000, 0x0196CA84) = 0
>> kaio(AIOWRITE, 396, 0xB82BC000, 8192, 0x6D822000, 0x01B543CC) = 0
>> kaio(AIOWRITE, 313, 0xB5ED4000, 8192, 0x1FCD2000, 0x019EE1D8) = 0
>> kaio(AIOWRITE, 395, 0xBF256000, 8192, 0x21AE6000, 0x01971958) = 0
>> kaio(AIOWRITE, 313, 0xB1700000, 8192, 0x1FCD4000, 0x019B65D0) = 0
>> kaio(AIOWRITE, 396, 0xBA238000, 8192, 0x24F86000, 0x01B5E32C) = 0
>> ....
>>
>>> So individual writes are synchronously, they are just fired off
>>> asynchronously.
>>>
>>> Anjo Kolk.
>>>
>>> claudio cutelli wrote:
>>>
>>>> Hi,
>>>> which background process had the problem?
>>>> because if lgwr, it already write in sync mode ...
>>>>
>>>> ----- Original Message -----
>>>> To: "Multiple recipients of list ORACLE-L" <ORACLE-L_at_fatcity.com>
>>>> Sent: Monday, April 15, 2002 5:33 PM
>>>>
>>>>> We have implemented a Sun Solaris Cluster (4 machines),
>>>>> connected
>>>>> to
>>>>> an EMC storage array. The migration began last fall, and we now
>>>>> have
>>>>> 15
>>>>> Oracle instances, with a mixture of 8.1.6 and 8.1.7, located there.
>>>>> We
>>>>> recently have had 2 occurances of asynchronous I/O wait times
>>>>> exceeded.
>>>>> When this occurs, every database crashes at the same time. The
>>>>> solution
>>>>> from EMC is to turn asynchronous I/O off in all of the Oracle
>>>>> instances
>>>>> (disk_async_io = false) and to increase the database writer slaves
>>>>> (dbwr_io_slaves = <something not 0>) to emulate asynchronous I/O.
>>>>> Has anyone run into this problem before? If so, how did you
>>>>> "correct" it? My feeling is that EMC is trying to give us a bandage
>>>>> to
>>>>> cover up the real problem, by trying to get Oracle to ignore it.
>>>>>
>>>>> Thank you.
>>>>>
>>>>> --
>>>>> Scott Canaan (srcdco_at_rit.edu)
>>>>> (585) 475-7886
>>>>> "Life is like a sewer, what you get out of it depends on what you
>>>>> put
>>>>> into it" - Tom Lehrer
>>>>>
>>>>>
>>>>> --
>>>>> Please see the official ORACLE-L FAQ: http://www.orafaq.com
>>>>> --
>>>>> Author: Scott Canaan
>>>>> INET: srcdco_at_ritvax.rit.edu
>>>>>
>>>>> Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
>>>>> San Diego, California -- Public Internet access / Mailing
>>>>> Lists
>>>>> --------------------------------------------------------------------
>>>>> To REMOVE yourself from this mailing list, send an E-Mail message
>>>>> to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
>>>>> the message BODY, include a line containing: UNSUB ORACLE-L
>>>>> (or the name of mailing list you want to be removed from). You may
>>>>> also send the HELP command for other information (like subscribing).
>>>>>
>>>> --
>>>> Please see the official ORACLE-L FAQ: http://www.orafaq.com
>>>> --
>>>> Author: claudio cutelli
>>>> INET: c_cutelli_at_hotmail.com
>>>>
>>>> Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
>>>> San Diego, California -- Public Internet access / Mailing
>>>> Lists
>>>> --------------------------------------------------------------------
>>>> To REMOVE yourself from this mailing list, send an E-Mail message
>>>> to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
>>>> the message BODY, include a line containing: UNSUB ORACLE-L
>>>> (or the name of mailing list you want to be removed from). You may
>>>> also send the HELP command for other information (like subscribing).
>>>
>>>
>>> --
>>> Please see the official ORACLE-L FAQ: http://www.orafaq.com
>>> --
>>> Author: Anjo Kolk
>>> INET: anjo_at_oraperf.com
>>>
>>> Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
>>> San Diego, California -- Public Internet access / Mailing Lists
>>> --------------------------------------------------------------------
>>> To REMOVE yourself from this mailing list, send an E-Mail message
>>> to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
>>> the message BODY, include a line containing: UNSUB ORACLE-L
>>> (or the name of mailing list you want to be removed from). You may
>>> also send the HELP command for other information (like subscribing).
>>>
>>>
>> // George Schlossnagle
>> // Principal Consultant
>> // OmniTI, Inc http://www.omniti.com
>> // (c) 301.343.6422 (e) george_at_omniti.com
>> // 1024D/1100A5A0 1370 F70A 9365 96C9 2F5E 56C2 B2B9 262F 1100 A5A0
>>
>> --
>> Please see the official ORACLE-L FAQ: http://www.orafaq.com
>> --
>> Author: George Schlossnagle
>> INET: george_at_omniti.com
>>
>> Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
>> San Diego, California -- Public Internet access / Mailing Lists
>> --------------------------------------------------------------------
>> To REMOVE yourself from this mailing list, send an E-Mail message
>> to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
>> the message BODY, include a line containing: UNSUB ORACLE-L
>> (or the name of mailing list you want to be removed from). You may
>> also send the HELP command for other information (like subscribing).
>
>
> --
> Please see the official ORACLE-L FAQ: http://www.orafaq.com
> --
> Author: Anjo Kolk
> INET: anjo_at_oraperf.com
>
> Fat City Network Services -- (858) 538-5051 FAX: (858) 538-5051
> San Diego, California -- Public Internet access / Mailing Lists
> --------------------------------------------------------------------
> To REMOVE yourself from this mailing list, send an E-Mail message
> to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
> the message BODY, include a line containing: UNSUB ORACLE-L
> (or the name of mailing list you want to be removed from). You may
> also send the HELP command for other information (like subscribing).
>
>

// George Schlossnagle
// Principal Consultant
// OmniTI, Inc 		http://www.omniti.com
// (c) 301.343.6422   (e) george_at_omniti.com
// 1024D/1100A5A0  1370 F70A 9365 96C9 2F5E 56C2 B2B9 262F 1100 A5A0

-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.com
-- 
Author: George Schlossnagle
  INET: george_at_omniti.com

Fat City Network Services    -- (858) 538-5051  FAX: (858) 538-5051
San Diego, California        -- Public Internet access / Mailing Lists
--------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).