jim.spath/blog/2008/09/05/stuck-in-the-update-part-1
Detective's log, Wednesday, September 10, 2008
In a stakeout, the operative tools, are patience, fortitude, and calmness. When looking for an intermittent software bug, these traits are also helpful. I'd also add experience, moxie, a dash of luck, and a certain, oh I don't know, rugged individualism? But watching for someone else to solve a case takes an entirely different skill set.
Yesterday, after chatting with the case officer, we sat back to wait for the answer. More precisely, we worked on other cases. Later in the day, we heard back. Not enough information; our sample wasn't complete; our evidence was faulty; the test user security was incorrect. So we supplied more description of what was wrong, when the last patches were applied, what was in the patches (standard SAP support package, we said).
Late yesterday, I posted an update, explaining again why I don't believe there is a database problem (see my CV here).
I also looked over their virtual shoulders, scanning the workload statistics for what transactions were being run, how long they took, and when they were on the job. Like looking at someone else's case history. Here are examples of what I saw:
So they started working early, for our timezone. At 4:11 I'm usually asleep, but not always. They're working in a different time zone, GMT or BST, or Eire.
ST12 I'm not that familiar with, but STAD is looking into the workload statistic repository. But if the problem transaction isn't exhibiting the same symptoms as we're reporting, looking at that information is a waste of time.
Unfortunately, it doesn't look like any work was done on our case all day.
So, on to the next day.
Detective's log, Thursday, September 11, 2008
I had a field trip on another case in the morning, and by the time I got back in the office I was behind on phone calls, email and other paperwork. So I was a little surprised when a Basis team member reported the case had been updated.
The news wasn't promising.
At 15:26:14 CET, the newest member of the support squad wearing out their gumshoes on this dilemma reported that the case was solved. It was a simple matter of "table fragmentation".
To an experienced DBA, those words are pretty close to an insult.
To a DBA emeritus, those words show a misunderstanding.
To an SAP mentor, let's just say, it's a "teachable moment."
Here's a sample of my reply:
The statements are similar to this:
SELECT WHERE "MANDT" = '040' AND "OBJEK" IN ( '0000000001' , '0000000001', '0000000001' , '0000000001' , '0000000001' ) AND "MAFID" = 'K'
I would like a developer to explain why the SAME EXACT value is searched for multiple times.
In closing, I asked:
"...you have not duplicated the problem conditions yet. If you would like one of us to execute the transaction for you to trace, please propose a time during US Eastern time zone working hours."
Let's see what develops tomorrow. The case has only been open with SAP for 2 weeks.
3 Comments
