From: Daniel Berger <Daniel.Berger@...>
Date: 2005-07-12T01:38:14+09:00
Subject: Strange logger rotation failure

Hi all,

Ruby 1.8.2
Solaris 10

I have a couple Ruby scripts running via cron that connect to an Oracle 
database and execute some stored procedures - nothing fancy.  On 
Saturday, July 2, 2005, the logger package spewed a strange error at me:

Your "cron" job on itwsp193910wss
log_cleanup_dev.rb

produced the following output:

/usr/local/lib/ruby/1.8/logger.rb:516:in `write': Shifting failed. 
'log_cleanup_dev.log.20050630' already exists. (Logger::ShiftingError)
	from /usr/local/lib/ruby/1.8/logger.rb:318:in `add'
	from /usr/local/lib/ruby/1.8/logger.rb:399:in `fatal'
	from log_cleanup_dev.rb:68

First, a bit of background.  There are two nearly identical scripts that 
run - one is log_cleanup.rb and the other is log_cleanup_dev.rb.  The 
only difference between them is the database connect string (i.e. one 
connects to a production database, and the script in question connects 
to a dev database).

The first curious thing to me is that this error was raised on July 
*2nd* (it also happened on July 8th, btw).  The second (less) curious 
thing is that this was only raised by the dev script.  It's less 
curious, because the reason the script failed in both instances was 
because the DBA was doing some maintenance on the dev database on those 
mornings.

# Relevant snippet from log_cleanup_dev.rb (line 68 noted):

# Logger instance
log = Logger.new("log_cleanup_dev.log", "monthly")

# and later...
begin
    sth = dbh.prepare("begin my_proc; end;")
    sth.execute
rescue DBI::Error => e
    msg = "Error executing stored procedure 'my_proc': #{e.errstr}"
    log.fatal{ msg } # Line 68
    raise
rescue => e
    msg = "Unexpected error executing stored procedure 'my_proc': #{e}"
    log.fatal{ msg }
    raise
ensure
    sth.finish if sth
end

Any ideas?  It looks to me like it's trying to rotate when it shouldn't 
or something, but I haven't dug too deeply into this yet.

Regards,

Dan