Re: Question about Archivemode

"Howard J. Rogers" wrote:
>
> And there's nothing special about ARCH. All that does is to automatically
> copy the online logs, which you can do yourself (as you've done here) or you
> can get a server process to do it (as happens with an alter system archive
> log all, for example).

Which got me thinking and testing again...

> But this is the most convincing demonstration that you can make your own
> guarantees in all sorts of ways that I've seen in a long, long time (and I
> intend to expropriate it mercilessly for teaching purposes).

And, in case anyone had any lingering doubts as to *why* running in ARCHIVELOG mode is the only guaranteed and recommended way of maintaining an unbroken stream of redo, have a look at this (try this a home kids, not on your production datacenter):

[This picks right back up after my previous test]

SEANTEST>
SEANTEST> alter system archive log current; alter system archive log current
*
ERROR at line 1:
ORA-00258: manual archiving in NOARCHIVELOG mode must identify log

[OK, so we'll just be specific then...]

SEANTEST> alter system archive log group 6   2 /

[And this is where one of my hard drives starts making all kinds of noise, much more than usual for such small redo logs... hmm... goes on for minutes... checking archivelog dest I see arch_1_1889.arc and arch_1_1888.arc, the former with a timestamp from a few minutes ago, the latter with a current timestamp. No matter how long I wait, I never get my prompt back from my archive current command, and the timestamp on 1888 stays current without the file getting any bigger. It's like it's stuck writing to that one. Time to check the alert.log:]

ARCH: Beginning to archive log# 6 seq# 1889
ARCH: Completed archiving log# 6 seq# 1889
ARCH: Beginning to archive log# 4 seq# 1888
ARCH: Completed archiving log# 4 seq# 1888
ARCH: Beginning to archive log# 4 seq# 1888
ARCH: Completed archiving log# 4 seq# 1888
ARCH: Beginning to archive log# 4 seq# 1888
ARCH: Completed archiving log# 4 seq# 1888
ARCH: Beginning to archive log# 4 seq# 1888
ARCH: Completed archiving log# 4 seq# 1888
ARCH: Beginning to archive log# 4 seq# 1888
ARCH: Completed archiving log# 4 seq# 1888
ARCH: Beginning to archive log# 4 seq# 1888
ARCH: Completed archiving log# 4 seq# 1888
ARCH: Beginning to archive log# 4 seq# 1888
ARCH: Completed archiving log# 4 seq# 1888
ARCH: Beginning to archive log# 4 seq# 1888
ARCH: Completed archiving log# 4 seq# 1888
ARCH: Beginning to archive log# 4 seq# 1888
ARCH: Completed archiving log# 4 seq# 1888
ARCH: Beginning to archive log# 4 seq# 1888
ARCH: Completed archiving log# 4 seq# 1888
ARCH: Beginning to archive log# 4 seq# 1888
ARCH: Completed archiving log# 4 seq# 1888

... ad infinitum

Yeah, I'd say it was stuck alright.

So I go and try to kill my session via another sqlplus window:

SEANTEST> alter system kill session '9,1'; alter system kill session '9,1'
*
ERROR at line 1:
ORA-00031: session marked for kill

but still nothing. Didn't feel like waiting around for SMON, so I get the PID and try a kill. Nope. Kill -9. Bingo! From the orignal screen:

SEANTEST> alter system archive log group 6   2 /
alter system archive log group 6
*
ERROR at line 1:
ORA-03113: end-of-file on communication channel

All of which is probably some sort of freak bug in 8.1.7.0.0, but I certainly wouldn't have the nerve to report it... ("well, you see, I wanted to archive an online redolog in a noarchive database so I can manually reinvent Oracle's built in archiving mechanisms and...").

Regards,
Sean Received on Thu Jul 18 2002 - 16:32:10 CDT