Oracle FAQ Your Portal to the Oracle Knowledge Grid

Home -> Community -> Mailing Lists -> Oracle-L -> RE: dbwr high count of threads

RE: dbwr high count of threads

From: David Miller <dm32840_at_bachelor.West.Sun.COM>
Date: Wed, 11 Sep 2002 10:11:30 -0500 (CDT)
Message-Id: <>

Hi Richard,

Ok. Since you are using filesystems (and I assume no directio on the ufs filesystem), and have not specified asynch_io, the default is to have it on. Because kaio (kernel asynch I/O) is not supported on ufs without directio or vxfs without QIO, Oracle will be using libaio (library asynch I/O) instead. The default number of worker threads in libaio is 256, so that's where your 258 threads count comes from. So things are working perfectly normally.

Besides on any modern system 258 threads is not a "high number".

Hope this helps.

Dave Miller

>From: "Ji, Richard" <>
>To: "'David Miller'" <>
>Subject: RE: dbwr high count of threads
>Date: Tue, 10 Sep 2002 16:49:52 -0400
>Sorry I didn't provide all the details.
>It's Oracle EE on Solaris 8.
>The development box is on UFS and Production on Veritas without QIO.
>Both are showing 258 LWP.
>I didn't specify ioslaves. Both db_writer_proesses and db_block_lru_latches
>are using default. Dev has 1 CPU, Prod has 2.
>"ps -edfL | grep dbw" shows 258 lwp under dbw0 process. Top also shows
>258 threads.
>No the database is not on NFS. We only used NFS to copy the datafiles
>during hotbackup. It was a server panic, due to NFS. Sun just confirmed it
>that it's a known NFS bug that under heavy usage, NFS could cause kernel
>I am looking into KAIO as this seems to be related to those LWP threads.
>I was just wondering I am seeing it on all of my databases and I didn't
>set anything special. Does other people see the samething on their plain
>vanilla database?
>-----Original Message-----
>From: David Miller [mailto:dm32840_at_bachelor.West.Sun.COM]
>Sent: Tuesday, September 10, 2002 3:09 PM
>Subject: Re: dbwr high count of threads
>Hi Richard,
>A couple of questions.
>What version of Oracle?
>What version of Solaris?
>Are you using asynch I/O?
>Are you on filesystems? If so, which one (ufs, vxfs, vxfs with quickio)?
>Have you specified ioslaves? If so, how many?
>How did you determine how many threads you were using?
>Is the entire database on NFS?
>Is the server crash a Solaris crash or and Oracle crash?
>Dave Miller
>>X-Unix-From: Mon Sep 9 16:26:57 2002
>>Date: Mon, 09 Sep 2002 14:18:24 -0800
>>To: Multiple recipients of list ORACLE-L <>
>>X-Comment: Oracle RDBMS Community Forum
>>X-Sender: "Ji, Richard" <>
>>From: "Ji, Richard" <>
>>Subject: dbwr high count of threads
>>X-ListServer: v1.0g, build 72; ListGuru (c) 1996-2001 Bruce A. Bergman
>>Mime-Version: 1.0
>>Content-Transfer-Encoding: 7bit
>>Hi all,
>>I noticed the dbwr process on Solaris has a very high number of threads
>>To me this is not a problem since I am seeing this on my small development
>>box too.
>>But we recently had some server crash and the consultant is saying Oracle
>>a lots of resource, citing the high number of Oracle thread count from the
>>core dump analysis.
>>I don't believe this lead to the crash because the core stack trace points
>>to NFS calls in both
>>However, I don't know how to explain the high number thread count mostly
>>from the dbwr process.
>>Is this normal? I mean, it looks like it's normal since I see this on all
>>of my instances.
>>How do I convince him that this is ok?
>>Thanks for your help.
>>Please see the official ORACLE-L FAQ:
>>Author: Ji, Richard
>> INET:
>>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: (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
Received on Wed Sep 11 2002 - 10:11:30 CDT

Original text of this message