Re: Filled up the DASD! Now can't run a specific query...

On 09/10/2015 02:13 PM, Bob Cagle wrote:

So I set Fix Central to download the latest CUME, Group, and TR PTFs directly to my system and left it to run for the evening, just to have it fill our remaining disk and crash the system.

I've since recovered from that fiasco and gotten the system back up and the utilization back down around 78% - it's a small 520, only 175gb total available. Will need to figure out how to get those PTFs loaded eventually.

After getting back to normal operations this morning, we discovered a specific report that just won't run. This is a JD Edwards World Writer - for those that don't know, it's very similar to Query, just specific to the JDE environment.

This report does a join on two specific PFs - simple report: join the files, then print the records where the price in one doesn't match the other. It's been running multiple times a day, every day, for years.

But now after the DASD issue, the job runs, but isn't doing anything I can see. It shows as active on WRKACTJOB, but no activity in the file I/O, call stack, or job log - it just sits there with no movement.

I thought I had it tracked down to a damaged join LF. I wasn't able to pull up this LF using DBU - just got a continual loading message, but could pull up each of its base PFs. So I recreated the join LF. No change.

Just now I've tried recreating this report with SQL in the iSeries Navigator Run SQL Scripts. Same thing - it's just sitting and spinning. Again, this isn't a fancy report and should only take a couple of minutes to run on our system with the data we have.

Any ideas on what's going on here? Is there an index that's been damaged/needs rebuilt?

I should note that as part of recovering space yesterday, I did a RCLSTG and an IPL. Now not sure if that was a help or a hindrance.

Thanks for any input,

Bob Cagle
IT Manager
Lynk

This may not be helpful, considering an IPL was done. But, curious, are
any object locks visible? Could the report be trying to acquire a
lock? Does report time out at some point?

Also what comes to mind from misadventures years ago, is content of a
data area inconsistent? That would be possible if a program needed to
see a value, but a retrieved value was not as expected - seen that too
many times.

Totally off the wall: I saw program code that requested operator to
enter one of two single letters as a response. Not unlike (C G). But,
code never tested for any other value. I spent time, out of morbid
curiosity, slogging through the program to see what happened. Gave up.
Just too twisted. But, the inspiration for a data area that does not
contain appropriate data. Did RCLSTG deposit anything related in the
QRCL (I think, maybe QRECOVERY) library related to J D Edwards?

John McKee