Re: Oracle Err: Can not allocate log, archive required

From: <hailey_kyle_at_my-deja.com>
Date: Wed, 31 Jan 2001 01:40:38 GMT
Message-ID: <957qem$806$1@nnrp1.deja.com>

There is a case I ran into recently where the archiver would stop, but its when you have a standby database. It doesn't sound like you'd be using a standby database, but its an interesting case and I've meant to post it as a warning to other. There is a bug, 1355687, where if the standby db dies in the middle of an archive, even though it is optional, the main db archiver will hang.

Interesting stuff to check is to see what the archiver is waiting for in v$session_wait. In our case it was waiting for 'file identify'. I wasn't here when it happened but captured it in monitoring scripts. The file identify wait's P2 field is called "file name". It is actually a pointer to the memory address in the archiver process where the name of the file it is trying to identify. On NT I'm not sure what you'd do, but on LINUX its easy to attach a debugger like gdb to the arc process and print that memory location and see what file it is waiting on.

In your case, during the hang, I'd look at CPU and disk io. Is anything running, is there disk IO, is the archiver process active or idle? In the case of an idle archiver during a hang, try checking out what the archive process is waiting for. Look at v$session_wait for SID = arc's SID.

In the case of blasting through the redo logs faster than the archiver, I wouldn't expect the archive wait to be very long. The archivers work is easier than the log writers.

Things you can do. Try

SQL> archive log all
what happens?

For more info and tracing you can try:

       SQL> alter system set LOG_ARCHIVE_TRACE=127;
       SQL> alter session set events 'immediate trace name systemstate
level 10'
       SQL> oradebug setorapid <pid of archiver>
       SQL> oradebug unlimit
       SQL> oradebug dump errorstack 3

to trace the database and archiver.

Best
Kyle

PS On the managed standby, more dangers, if the standby database has an Oracle 600 it will bring the primary database down ! (fixed in 8.1.7 bug 956434)

In article <9574g3$jfl$1_at_nnrp1.deja.com>, alexberi_at_my-deja.com wrote:
>
>
> I have configured an Oracle NT database 8.0.5 to work in archieve mode
> and set automatic archieving of logs, but I receive the message:
>
> "Can not allocate log, archive required"
>
> There is plenty of room on the drive where the archive files are
> created, and I have about 100 archieved logs there.
>
> Can anyone tell me where is the problem?
>
> Thanks, Alex.
>
> Sent via Deja.com
> http://www.deja.com/
>

Sent via Deja.com
http://www.deja.com/ Received on Tue Jan 30 2001 - 19:40:38 CST