|
Full
Sets and Inducing Sleep Inside VESOFT covers
tips and techniques you can use with VESOFTs products,
especially MPEX. So when I assumed responsibility for a system that did nightly processing and involved heavy updating of several databases, I had to figure out a way to avoid those pesky calls at hours that had single digits. This was in the Mesozoic Era when disc was still expensive, so we could not just set the capacities as high as we wanted and ignore it. We had to manage it. I hate trying to remember to do something every day, so I needed to figure out something that would happen on its own and let me know if theres a problem. Enter MPEX and the DBSETFULLNESS file attribute. A regularly scheduled job with the command: %LISTF @.GROUP.ACCOUNT(DBSETFULLNESS>.9) This would find any datasets more than 90 percent full. Eventually when that thing called e-mail came into common use, I added a nice step that would send an e-mail to tell me about any problems. I had an almost foolproof means of avoiding the DATASET FULL phone call. (Of course, nothing is foolproof, as Ill explain in the next Solving a mystery section.) There are several other MPEX attributes out there for datasets. Need to find all your automatic masters? Try %LISTF @.@.@(DBSETTYPE=A). (The other choices there are D and M.) You can search for datasets by name (DBSETNAME), number of entries (DBSETENTRIES), high-water mark (DBSETHIGHWATERMARK), blocking factor (DBSETBLOCKFACTOR), numbers of paths (DBSETNUMPATHS) and even wasted space in a block (DBSETBLOCKWASTAGE). And those file attributes are not just for simple files and database-related files you can use MPEX to select spool files by specific criteria too. The attribute always begins with SPOOL. followed by the specific item to look for. SPOOL.ISACTIVE finds any active spool files (theres our old friend IS). Other booleans to tag along with SPOOL. are ISREADY, ISOPENED, ISLOCKED, ISDEFERRED and ISSPSAVE. You can also look for attributes like the JOBNUMBER, the USER, the ACCOUNT, the FILE (actual file name), the SPOOLFILENUM and the DEVICE. If theres a forms message, then HASFORMSMSG is what you are looking for. Now, you might ask why anyone would use this, but I have a real life example. Once a year, I had to convert about 300 reports from spool file to PDF. To do the conversion I used OpenSeas Fantasia (my favorite print formatter and an excellent product). Well, I didnt want to do this manually 300 times, so I set up a REPEAT...FORFILES command file. I had all the reports spooled out with an output priority of 2 (a priority we generally never use). Then I would run the following MPEX command file: repeat run laser.laser.larc;parm=97; <other Fantasia parameters> forfiles @.out.hpspool(spool.outpri=2) The command file looks for every spool file with the priority of 2 as represented with the @.out.hpspool(spool.outpri=2) and it would run them through Fantasia. It worked like a charm, and yet again MPEX made my life much easier. Solving a mystery When I was playing with examples for this column, I created the job at right. It looked for any datasets fuller than 80 percent. (Figure 1 below shows the edited form, with its output edited for brevity). Figure 1 JOB DBCHECK,MANAGER.XXXXXXX. Priority = DS; Inpri = 8; Time = UNLIMITED seconds. Job number = #j290. THU, JAN 3, 2002, 11:50 AM. HP3000 Release: C.60.00 User Version: C.60.00 MPE/iX HP31900 C.16.01 Copyright Hewlett-Packard 1987. All rights reserved. STREAMED BY STEVE,MANAGER.XXXXXXX(#S192) ON LDEV# 266 STREAM DATE: THU, JAN 3, 2002, 11:50 AM :comment ******** :comment * this job checks for datasets > 80% full :comment * in the shared group :comment ******** :run main.pub.vesoft;parm=1 MPEX/3000 29N00420 (c) VESOFT Inc, 1980 6.0 03:06927 For help type 'HELP' %listf @.shared(DBSETFULLNESS>.8),DB MPEX %LISTF @.shared(DBSETFULLNESS>.8) PAGE 1 DBCHECK,MANAGER.XXXXXXX,PUB THU, JAN 3, 2002, 11:50 AM ACCOUNT= XXXXXXX GROUP= SHARED FILENAME SET NAME TYPE LDEV ENTRY CAPA- %FULL COUNT CITY FWRTBL02 TAX-YEAR-MAST M 35 2 2 100.0% HISTDB06 HIST-BIO D 31 548129 755157 72.6% HISTDB07 HIST-MCAT77 D 32 640632 773000 82.9% HISTDB08 HIST-MCAT91 D 45 124859 163728 76.3% IRRGDB01 IRREG M 44 3287 3500 93.9% PULLDB02 APPLFILE D 30 1815 2004 90.6% REFRDB07 SCHOOL-KEY A 45 209 250 83.6% %exit Heres the mystery. Correct me if I am wrong, but when I learned percentages in fourth grade, 72.6 percent was less than 80 percent. Then why would MPEX tell me that two datasets that are about 75 percent full are greater than 80 percent full? Now, my dear Watson, I can certify that the people at Vesoft are smart people, so this is not a stupid error. There must be something were not seeing. In fact, the command is correct, and Holmes is also correct. Were just not seeing all the numbers. The key is that both those datasets use DDX (dynamic dataset expansion). In the first case, the entry count is correct, but the MAXIMUM capacity is shown, not the current capacity, which is 127656. Divide the values and we get 97.8 percent! Now thats more than 80 percent. The second item is the same with a fullness of 91.3 percent. So we see that the DBSETFULLNESS value is determined by dividing the entry count by the current capacity, but the LISTF,DB displays the max capacity and computes the %FULL using the max and the entry count.
Mystery solved! Copyright The 3000 NewsWire. All rights reserved. |