ARCH Process and ALERT Log

[Sameer]

Hello,

My database is running in ARCHIVELOG mode. I have set parameter log_archive_start to TRUE for automatic archiving of redolog files. It works fine

But sometimes I get error in my ALERT log file saying

ARC0: Failed to archive log# 1 seq# nNumber

or

ARC1: Failed to archive log# 1 seq# nNumber

But I also get

ARC0: Beginning to archive log# 2 seq# 386
Mon Jul 29 16:55:53 2002
Thread 1 advanced to log sequence 388
Current log# 1 seq# 388 mem# 0: D:\ORACLE\ORADATA\ORCL\REDO01.LOG
Mon Jul 29 16:55:53 2002
ARC0: Completed archiving log# 2 seq# 386

and same for ARC1 also

Will anybody let me know ... why my ARCH process fails in between?

1. How do I findout how many ARCH processes are running?
2. How to start multiple ARCH process?


Thanks

Sameer

[SANJAY_G]

How many redo log groups do you have ?
Try addding one or two more log groups to avoid contention.

Sanjay

[moff]

Hi,

Are your redo logs and Archive logs on the same disk? If so move them to seperate disks.
I think the parameter log_archive_max_processes can be used to define more Arch processes (but I would check this beofre setting it).

If your on Unix you can see how many arch processes are running by entering the following at the command line...

ps -ef |grep arc

These are information messages and as long as they're followed by a 'Completed Archiving' message for that log file should be OK. Look at ways of tuning the Arc process. (you could try http://www.ixora.com.au/tips/tuning/archiver.htm for info)

Regards
Moff.

[Edited by moff on 07-30-2002 at 05:53 AM]

[Sameer]

Thanks for replies..

Sanjay: I have three logfiles.. Do you mean .. I should try adding a member to log group or adding a new logfile?

Moff: Yes.. My redo logs and Archive logs are on the same disk..

Well.. This is not a PROD database.. so don't wonder why these are on the same disks..

I will try relocating ARCH logs on different drive.

Thanks

Sameer

[Tarry]

Since you have 3 loggruops you don't need to add any more.(well you can always ) but i don't think that's necessary.
Destination of arch files.
try to reduce the time of redo log switching by increasing the size of redo logfiles, on a default db creation it's 1mb each ,make it 10mb each or so.
you don't need to create any aditional arc processes as the lgwr automatically starts one if necesary but you can do it if you want to.

[TomazZ]

Sou you have (at least) two archive processes: ARC0, ARC1.

I noticed that sometimes two archive processes are competing for same log file.

Can you find in alert log, when one of those two (for instance: ARC0) fails, somewhere near you can have message that other (ARC1) succesfully archived the same log sequence number.

If so, this is 'normal'
Well, nothing to really worry about, i think. Anyway, you can restart archiver, then you should have only one.

Tomaz

[Sameer]

Hey.. You are right...

See below

Thu Jul 18 11:36:49 2002
ARC1: Beginning to archive log# 2 seq# 230
Thu Jul 18 11:36:51 2002
Thread 1 advanced to log sequence 232
Current log# 1 seq# 232 mem# 0: D:\ORACLE\ORADATA\ORCL\REDO01.LOG
Thu Jul 18 11:36:51 2002
ARC0: Beginning to archive log# 2 seq# 230
ARC0: Failed to archive log# 2 seq# 230
ARC0: Beginning to archive log# 3 seq# 231
Thu Jul 18 11:36:51 2002
ARC1: Completed archiving log# 2 seq# 230
ARC1: Beginning to archive log# 3 seq# 231
ARC1: Failed to archive log# 3 seq# 231


ARC1 has started archiving before ARC0 (log# 2 seq# 230) so ARC0 fails..

later ARC0 starts archiving log# 3 seq# 231 before ARC1, then ARC1 fails to archive (log# 3 seq# 231)

Thanks.. But then why these processes doesn't know which log has already archived or in the process of getting ARChed by other archiver?

Thanks

Sameer

[TomazZ]

I don't know.
Probably ARC1 doesn't know that ARC0 is doing this log until it's finished, and just tries.
But it looks stupid...

Everything I found on metalink was explanation, that one is manually issued with 'alter system archive log current'.
But in my case it wasn't it. Nor in yours, i guess.

I can't really remember now, if restarting just archiver helps, or database should be rebounced. Try.

Have fun,
Tomaz

[SANJAY_G]

Does any of your hotbackup script issue-

ALTER SYSTEM ARCHIVE LOG ALL;

?


Sanjay

[chao_ping]

Originally posted by TomazZ
I don't know.
Probably ARC1 doesn't know that ARC0 is doing this log until it's finished, and just tries.
But it looks stupid...

Everything I found on metalink was explanation, that one is manually issued with 'alter system archive log current'.
But in my case it wasn't it. Nor in yours, i guess.

I can't really remember now, if restarting just archiver helps, or database should be rebounced. Try.

Have fun,
Tomaz

hi,
look at asktom.oracle.com, and you will know the answer, i post his words here: not his original words:
Just as tomazz said, one archiver process doesnot know the others is process(archiving) this log file, so there is error message like that, in fact, this is just a harmless warning, in 9i , it is improved.