HP3000-L Archives

March 1998, Week 2

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Jeff Kell <[log in to unmask]>
Reply To:
Jeff Kell <[log in to unmask]>
Date:
Mon, 9 Mar 1998 20:51:24 -0500
Content-Type:
text/plain
Parts/Attachments:
text/plain (34 lines)
Steve Donnelly wrote:
> re:  Using the User Logging subsystem.

> I forwarded your suggestion to use the logging facility to Rick
> Ingham, who is one of our programming supervisors.  It's a creative
> alternative, but will most likely require some work to implement.

(Rick's concerns snipped).

Another alternative, depending on the "latency" of your application
programs (i.e., how long are the logfiles open by individual programs)
here's another approach:

* create the files as plain flat files.
* "periodically" run a job which checks the EOF (if you're worried
  about space) or on a set schedule (if worried about time) which can
  switch the logfile via a Posix trick:
      mv logfile,logfile.[some_unique_suffix]
      callci build logfile;[whatever parms necessary]

Currently executing programs continue to write to the renamed file but
new programs will write to the newly built file.  There is granted a
window for error there (if something tries to open the logfile between
the mv and the build) but it is relatively small.

You can then post-process the logfiles as you wish.  The only immediate
drawback is that you don't have a contiguous, chronological record of
the transactions unless you have a timestamp in the files that you can
merge them together based on the time.

Very low overhead, no programming changes.

Jeff Kell <[log in to unmask]>

ATOM RSS1 RSS2