RE: Why does it take so long to write archive logs

From: John Carlson <jcarlson_at_cj.com>
Date: Tue, 08 Jul 2003 14:20:55 -0700
Message-ID: <F001.005C3E70.20030708135925@fatcity.com>

Oracle says it is not their problem. They are waiting for the confirmation that the file is written. It is written just like any sequential file. It does not pre-allocate and then load the data.

Yes, the database would stop after it had rolled through all log groups and could not continue until the oldest one finished archiving. We would see where other logs would archive just fine while one was taking forever. We have 8 log groups. So, for example, it would archive 1,2,and 3 then 4 would just sit there while it archived 5,6,7,8,1,2,3 and then the db would hang. It has taken over an hour for it to free up sometimes. This happened where the two machines were sitting side by side with a GigE connection between them writing to local disks as well as to a remote machine with a NetApp storage connection. It didn't seem to matter what the device was, just that it was archiving to another machine. As a work around, we turned off automatic archiving and used scripts to copy the archive logs. And yes, we had them set as optional and I believe that is why it let us continue to use up the other logs but once it wrapped around it could not continue.

John

-----Original Message-----
Sent: Tuesday, July 08, 2003 10:25 AM
To: Multiple recipients of list ORACLE-L

John,

Did Oracle give you any idea as to what Oracle was doing to the archive log files, while it was waiting for the hardware/network issues to be resolved?

I understand that the redo log information is written the archive log file, but does Oracle create a file with the correct size allocated and then put the information into it or does it create the file, write all the information into it and then update a byte within the file with a check digit to ensure the file is complete?

TIA, Bryan

-----Original Message-----
Sent: Monday, July 07, 2003 7:34 PM
To: Multiple recipients of list ORACLE-L

I had a similar problem on 9.2 and, just as you describe, I could watch the file on the remote location and the byte count would show the whole file was there but the alert log file would not show the archive complete for as long an one hour. When it did finally complete, the byte count on the remote file did not change. I worked with Oracle for several month before they finally convinced me it was a network problem. I do not really know anything about networks so I cannot really help you but our Unix admin guys made some changes in the network and the problem went away. Sorry, I cannot give you more info because we were also in the process of changing out a lot of hardware including network hardware so there is no magic parameter I can tell you to change. Just talk to your network gurus and hopefully, they can figure it out.

HTH,
John

-----Original Message-----
Sent: Monday, July 07, 2003 3:34 PM
To: Multiple recipients of list ORACLE-L

Hello everyone,

We have an 8.1.7.4 on HP-UX 11.0 OPS database that uses 200mb archive logs with 3 archive log destinations, the first destination is local to the machine which is mandatory, the second destination is a filesystem accessible via nfs which is optional and the third destination is a remote standby database accessible via a vpn which is also optional. We have 10 archive processes to take care of the writing of the archives. Both the local destination and nfs mounted filesystems are on a HP XP256 storage device.

This morning there was a process running that would update a table that is used for a catalog of parts. The process was producing archive logs, but the archives were taking around 10 minutes to write. During the process we would see the file created with the expected byte count of the file within the first minute, but the timestamp of the file would indicate access for the next ten minutes.

We have two other 8.1.7 databases that are setup in a similar manner (that are not OPS) that have 2 destinations, one local and the other being sent to a remote standy database. We are not seeing the same type of issue with the archive logs.

Can someone point me in the direction of either information or advice about why the archives may be taking so long to write?

Is it that we have 3 destinations and it waits until all 3 destinations are taken care of?

Why does the archive log file appear to be modified even though the byte count hasn't changed?

TIA, Bryan Rodrigues
Elcom, Inc.
Oracle DBA
--

Please see the official ORACLE-L FAQ: http://www.orafaq.net
--

Author: Rodrigues, Bryan
INET: [EMAIL PROTECTED]

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------

To REMOVE yourself from this mailing list, send an E-Mail message to: [EMAIL PROTECTED] (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.net
--

Author: John Carlson
INET: [EMAIL PROTECTED]

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------