Message-Id: <10600.115592@fatcity.com> From: "Danut Bancea" Date: Fri, 25 Aug 2000 15:18:41 -0400 Subject: Re: checkpoint error This is a multi-part message in MIME format. ------=_NextPart_000_028D_01C00EA7.C0742BA0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable OEMHi, In this case you have two way: 1) increase the size of redo logs file 2) add more redo logs files Or a combination of both. 3) You have also a third option to change all indexes for the tables = which are updated/inserted to nologging.. If you need to restore the database you need just to rebuild the = indexes.=20 Danut, ------------------------------------------------------------- Danut Bancea Oracle DBA Tele-Pages Inc.=20 Phone : (416) 296-9011 Ext. 2271 Fax : (905) 477-7062 E-mail : danut@telepages.ca ------------------------------------------------------------- ----- Original Message -----=20 From: Zeliha Y.D.=20 To: Multiple recipients of list ORACLE-L=20 Sent: Friday, August 25, 2000 3:51 AM Subject: Re: checkpoint error Hi Danut, the database is running in archive logs, the archive process can copy = fastly the redo logs. when I query v$log, I see ARCHIVED is "YES", but STATUS is "ACTIVE". sometimes, when running batch processes on database , this problem = appears. because of the completeness of the process, commit statement is at the = end of process. So that the commit statement is not too often. Zeliha YILMAZ DBA (ASKI) Tel : +903123115361 Fax : +903123241598=20 ----- Original Message -----=20 From: Danut Bancea=20 To: Multiple recipients of list ORACLE-L=20 Sent: Thursday, August 24, 2000 9:49 PM Subject: Re: checkpoint error Hi, I met with this problem before, and I found that the problem exits = only when the database is not in archive logs, or the archive process = can not copy fast enough the redo logs. Do you have the archive redo logs on separate hard drive ? How often do you commit during the mass insert/update ? Danut, ------------------------------------------------------------- Danut Bancea Oracle DBA Tele-Pages Inc.=20 Phone : (416) 296-9011 Ext. 2271 Fax : (905) 477-7062 E-mail : danut@telepages.ca ------------------------------------------------------------- ----- Original Message -----=20 From: Zeliha Y.D.=20 To: Multiple recipients of list ORACLE-L=20 Sent: Thursday, August 24, 2000 8:34 AM Subject: checkpoint error Hi all, Our system DUNIX 4.0D Alpha 4100 with 2 GB RAM, 2 x 467 MHZ CPU Oracle 7.2.3 I have a problem. I have 8 RedoLog files each 83 MB(Total 650 MB). In my system, sometimes Oracle hangs, at that time I got the following = ; SQL> select group#,status from v$log; GROUP# STATUS ---------- ---------------- 11 ACTIVE 12 ACTIVE 13 ACTIVE 14 ACTIVE 15 ACTIVE 16 CURRENT 17 ACTIVE 18 ACTIVE .. and in the alert log file I had the following... contents of alert.log: Thread 1 cannot allocate new log, sequence 8230 Checkpoint not complete Current log# 12 seq# 8229 mem# 0: /ctl/redo/log12SEVDE.rdo =20 The reason for this, there is a very large Update/delete/insert = transactions from 1 (or more) process. Someone offered to increase number of DBWR background process. how to do it? thanks,=20 zeliha. ------=_NextPart_000_028D_01C00EA7.C0742BA0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable OEM
Hi,
 
In this case you have two way:
 
1) increase the size of redo logs file
 
2) add more redo logs files
 
Or a combination of both.
 
3) You have also a third option to change all = indexes for the=20 tables which are updated/inserted to nologging..
 
 If you need to restore the database you need = just to=20 rebuild the indexes.
 
 
Danut,
 
 
-------------------------------------------------------------
Dan= ut=20 Bancea
Oracle DBA
 
Tele-Pages Inc.
Phone : (416) 296-9011 Ext. 2271
Fax : (905) = 477-7062
E-mail : danut@telepages.ca
 
-------------------------------------------------------------
----- Original Message -----
From:=20 Zeliha Y.D.
To: Multiple recipients of list ORACLE-L =
Sent: Friday, August 25, 2000 = 3:51=20 AM
Subject: Re: checkpoint = error

Hi Danut,
 
the database is running in = archive=20 logs, the archive process can copy fastly the redo = logs.
when I query v$log, I see = ARCHİVED is "YES", but=20 STATUS is "ACTIVE".
 
sometimes, when running batch = processes on=20 database , this problem appears.
because of the completeness of the = process,=20 commit statement is at the end of process.
So that  the commit = statement is not=20 too  often.
 
Zeliha YILMAZ
DBA (ASKI)
 
Tel  = : +903123115361
Fax : = +903123241598 
----- Original Message -----
From:=20 Danut=20 Bancea
To: Multiple=20 recipients of list ORACLE-L
Sent: Thursday, August 24, = 2000 9:49=20 PM
Subject: Re: checkpoint = error

Hi,
 
I met with this problem before, and I found that = the=20 problem exits only when the database is not in archive logs, or the = archive=20 process can not copy fast enough the redo logs.
 
Do you have the archive redo logs on separate = hard drive=20 ?
 
How often do you commit during the mass = insert/update=20 ?
 
 
Danut,
 
=
-------------------------------------------------------------
Dan= ut=20 Bancea
Oracle DBA
 
Tele-Pages Inc.
Phone : (416) 296-9011 Ext. 2271
Fax : = (905)=20 477-7062
E-mail : danut@telepages.ca
 
=
-------------------------------------------------------------
----- Original Message ----- =
From:=20 Zeliha Y.D.
To: Multiple=20 recipients of list ORACLE-L
Sent: Thursday, August 24, = 2000 8:34=20 AM
Subject: checkpoint = error

Hi all,

  Our = system
  DUNIX=20 4.0D
  Alpha 4100 with 2 GB RAM, 2 x 467 MHZ CPU
  = Oracle=20 7.2.3
  I have a = problem.
  I=20 have 8 RedoLog files each 83 MB(Total 650 MB). In my
  = system,=20 sometimes Oracle hangs, at that time I got the following=20 ;

 SQL> select group#,status from=20 v$log;

     GROUP# = STATUS
 ----------=20 = ----------------
         11=20 ACTIVE
         12=20 ACTIVE
         13=20 ACTIVE
         14=20 ACTIVE
         15=20 ACTIVE
         16=20 CURRENT
         17=20 ACTIVE
         18=20 ACTIVE

.. and in the alert log file I had the=20 following...

 contents of alert.log:

  Thread 1 cannot = allocate new log,=20 sequence 8230
  Checkpoint not = complete
   =20 Current log# 12 seq# 8229 mem# 0: /ctl/redo/log12SEVDE.rdo

 
 The reason for this, there = is a very=20 large Update/delete/insert transactions
from 1 (or more)=20 process.
 
Someone offered to increase = number of=20 DBWR background process.
 
how to do it?