RGZPFPCT - RGZPFM by percentage deleted records

[kitvb1]

This comes from www.support.changefit.com (source can be downloaded there if one registers. Registration is free.)

This command will perform a RGZPFM for each file (table) found matching the input criteria. The input parameters are soft, i.e. generic (*) names can be used. Allows for a percentage deleted records threshold to be input (New feature).

Important:
1. Member locks are catered for (new feature).
2. The current library/file/member being re-orged is displayed (new feature).
3. No footprint is retained after execution.
4. May be executed in batch or interactively.
5. After restoring on your box, use CHGCMD CMD(RGZPFPCT) PGM(xxxxx/RGZPFPCTC) where xxxx = the library to which you restored.

Objects created to v510. Tested from v530 thru v540.

Contents of ZIP file is a savf containing the CMD, CLLE, and RPGLE. Hope you will find it useful.

[DeadManWalks]

Re: RGZPFPCT - RGZPFM by percentage deleted records

Hi kitvb1, I went and looked at the site you mentioned and I'm interested in finding some info for the change management solution. I can see where it is on the site or the web.

Thanks.

[bryce4president]

Re: RGZPFPCT - RGZPFM by percentage deleted records

So is this reorg command faster than the standard reorg command?

[kitvb1]

Re: RGZPFPCT - RGZPFM by percentage deleted records

No, it's not. When I got to this site, we were running between 90% and 95% continually. So I wrote a little CL that would cleanup everything in the *ALLUSER libraries with >10% (hardcoded) deleted records. (Some file had 293 recs, but 3,5mil deleted recs.) As it was an emergency at the time (we were at 98%) i just let it time out which wastes a bit of time. But we still run it to this day as a weekend job.

So after the posts last week here, I rewrote the RPG to check for mbr locks and bypass that mbr. (This was a good learning curve for me 'cos I hadn't used a list API before.)

I also added the CMD wrapper so that one could specify the no. of deleted records threshold and allow more library options.

You should also note that there are more options allowed on the Library parameter as is allowed on the normal RGZPFM command.

[bryce4president]

Re: RGZPFPCT - RGZPFM by percentage deleted records

So you are sacrificing performance for flexibility with the new command? I just want to make sure I understand, that's all. I like the options you've created. How much of a performance hit would you expect to see?

[kitvb1]

Re: RGZPFPCT - RGZPFM by percentage deleted records

very little, it uses the standard RGZPFM to do the actual re-org... but a huge saving in time over the old version as it's no longer waiting for the system to reply (30 secs) that it can't re-org due to a mbr lock.

The only extra time taken is the call to the lock API for each member that is above the threshold.

[kitvb1]

Re: RGZPFPCT - RGZPFM by percentage deleted records

So is this reorg command faster than the standard reorg command?

Thought i'll just give you some feedback. I replaced the old version with the new version this weekend. The old version runs for an average of 2hrs 17mins (for May)

RGZPFPCT runs for 42mins.

15/06/08  2:31:45  000192091   QS
15/06/08  7:02:21  000192091   SS
15/06/08  7:02:22  000192091   RI
15/06/08  7:44:02  000192091   RS

QS=Queued Successfully, SS=Submitted Successfully, RI=Run Initiated, RS=Run Successfully

Quite a huge saving!!

ps. We have over 300 000 pf's and 60% of the runtime is just the dspobjd (*allusr/*all).

[kitvb1]

Re: RGZPFPCT - RGZPFM by percentage deleted records

Updated this command to cater for 0-100% deleted records (previously 1-100%).
Reason: RGZPFM may need to be run over a file containing only a few deleted records to correct an access path.

Please change appropriate line in the CMD source as follows:

            PARM       KWD(PCT) TYPE(*DEC) LEN(3) DFT(10) RANGE(0 +

[gcraill]

Re: RGZPFPCT - RGZPFM by percentage deleted records

Freaky - I am just doing our daily reorgs (Via MIMIX) and you post this up.

MIMIX program allows reorg while active, and it tracks journal changes and updates those as well, and then MIMIX promoter bangs in the new object when the reorg is completed. But it does NOT cater for member locks ... You got em all beat up there Kit!