Three months of SYSLOG sounds rather excessive, unless the user is
really willing to foot the bill for the extra JES SPOOL space, and the
additional cost of backing up all that data for DR. We only keep about
3 weeks of SYSLOG online to JES and just that much is typically 5 - 10%
of our total SPOOL space.
Be sure you are segmenting off SYSLOG data (specify lines per segment in
CONSOLxx ?) so that you typically get several segments per day. This
makes it possible to use standard JES Purge commands (potentially from a
scheduled batch job) to delete those segments that are older than "x"
days, while the SDSF LOG function will still see all segments as a
single SYSLOG file.
Rather than attempt the scheme you describe, I would use a Sunday
scheduled job with batch REXX to generate a sequence of commands with
appropriate date values for a batch execution of SDSF including a "PT
00.00.00 mm/dd/yyyy 00.00.00 mm/dd/yyyy" with suitable dates to offload
the previous week's SYSLOG data to DD that points to a GDG dataset, and
then issue the JES2 purge command to delete SYSLOG segments older than
"x" days, for "x" of your choice but greater than a week. At this point
the most recent "x" days of data will both be in some offloaded GDG and
in the still online JES2 SYSLOG segments. From SDSF all the remaining
segments may be viewed as if they were a single time-ordered entity and
the SDSF "locate" command can be used to quickly locate any date-time
point within the remaining portion of SYSLOG. By offloading the most
recent week each week instead of the oldest week, you don't have to
worry about the imprecision in the point at which old SYSLOG data is
purged (JES will purge at a segment boundary, which will not be at a
consistent time from week to week).
This should give the user convenient SDSF-accessible version of SYSLOG
for the last "x" days plus GDG archives of all weekly SYSLOGs up to the
GDG limit. No need to play games with copying remaining records, etc.
This ought to satisfy the user's requirement with minimal cost.
[EMAIL PROTECTED] wrote:
Hi All;
I have an unusual request from my user. He wants to keep three months of
SYSLOG on the spool in one logically contiguous piece and every week, he
wants to dump the oldest week to a GDG, leaving the rest there in a single
spool dataset. It's rather like a logger scenario. I have never heard of
anyone wanting this, let alone implementing it but, hey, that's what makes
life so interesting.
One could imagine a program that does this, it would have to read and write
the oldest week's records to the GDG and then copy the remaining records to
the start of the file and reset the end of file marker, all without JES
worrying his pretty little head about it. Fairly tricky I should think.
Any thoughts/ideas/droll comments would be much appreciated.
Thanks,
Jon Bathmaker
Senior Systems Programmer,
Cornerstone Systems, Inc.
2 Robert Speck Pkwy, 280
Mississauga, ON L4Z-1H8
Desk: 905-854-4156
Cell: 647-400-4156
Office: 905-275-9977
--
Joel C. Ewing, Fort Smith, AR [EMAIL PROTECTED]
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
