archived log stop working

Printable View

10-29-2001, 04:22 PM
mike73

My database was running archive log mode and all the sudden it stop archiving. Below is the errors on archiving.

ARC2: i/o error 19502 archiving log 4 to 'I:PRODDW\ARCHIVE\ARC49549.001'
ARC2: Archiving not possible: error exceed
ARC2: Failed to archive log# 4 seq # 49549

at some point later, I found out from this message.

Archiver process Freed from errors. no longer stopped

Why did archiver stop achived ??? and why it freed from errors ??? and start working again.

Could someone tell me why ???

I am running 8.1.6 on windows 2000

thanks

[Edited by mike73 on 10-29-2001 at 03:32 PM]
10-30-2001, 11:32 AM
mike73

No Reply !!!!!!!!!!

Please give me some feed back

Thanks
10-30-2001, 12:04 PM
marist89

Did I: have space when it stopped archiving? My bet would be that you ran out of space on I:. When somebody realized this or the backup kicked in, they started deleting files. When space freed up, the archiver kicked in again.
10-30-2001, 12:06 PM
amar

You most probably exceeded the max number of open files limit at the OS level temporarily.

Check out Bug# 1068882 fixed in 8.1.6.3

-amar
10-30-2001, 12:33 PM
snowtree

You could backup the old archive log file to tape or other drive, then delete them.
10-30-2001, 03:05 PM
Fizzlewhif

If you ran out of disk space you would have seen the message "O/S-Error: (OS 112) There is not enough free space on the disk" in your [sid]ARCH.TRC file as I recently had this happen. Your whole database would have stopped as well until free space was allocated.
09-17-2003, 10:24 PM
yls177

Quote:

Originally posted by marist89
Did I: have space when it stopped archiving? My bet would be that you ran out of space on I:. When somebody realized this or the backup kicked in, they started deleting files. When space freed up, the archiver kicked in again.

right.. i think its not the space issue.. .but rather the IO error... could be the disk.. but i am not sure..

cos when its the space issue i will get the below

Quote:

If you ran out of disk space you would have seen the message "O/S-Error: (OS 112) There is not enough free space on the disk" in your [sid]ARCH.TRC file as I recently had this happen. Your whole database would have stopped as well until free space was allocated.
09-18-2003, 03:23 AM
vishal sood

The same problem faced by me few days back. It is because of less space on Archive disk. When u free the space then archiveing will start again autometically.

I question may arise why the Database not hanged ? when archive process fails.

The answar is

Suppose u have 3 Log groups.

when log swtich occurs and LGWR starts writing next online Redo log file, and the archive process do the archive of privios log file. On next log swtich LGWR starts writing next on line group. and Arch. process keep on tring archive log files and found no space and starts issuing archive related errors. The database will hang when all the 3 Log groups are full and archive is not done and no log group is available for Redo entries.
09-18-2003, 03:51 AM
wengyan

yes.I agree vishal sood.
Oracle database will hang whenever not log group can been rewrite.
09-18-2003, 04:41 AM
DaPi

I think I can only add to the problem! anyway, here is my evidence:

This is EE 8.1.7.4.10 under Windows NT4.0 Server SP6a

I had the same this Monday morning at 04:50, the previous log switch was on the Saturday.

1) What caused the log switch at 04:50? - no users, no batch process - can checkpointing do this?
2) The OS/Hardware/RAID logs show no errors - this is a RAID-1 pair.
3) I'm running 8.1.7 - so it shouldn't be the Oracle bug.
3) I expect that 10 minutes later an independant OS process then moved the archlog file to another directory (same disk) and thus allowed ARCH to write a second version of the archlog.
4) The first version of the archlog that was written fails when applied to the Standby with "ORA-00317: file type 0 in header is not log file", the second works fine.
5) There was 1.3Gb free on the destination disk when I got in at 08:30. As ARCH could write a second version 10 minutes after the first, it could not be a space problem.
6) I can't imagine it was due to hitting an OS limit at that time of day.

This leaves me with the conclusion of data corruption caused by an Act of God, undetected at the OS/Hardware level.

added later:
7) No related *.trc files