Toad World® Forums

Trace file browser inconsistency (doubts)

Hi

Toad 12.5.0.50 x64, Oracle Client 11.2.0.4 against Oracle database 11.2.02

When I open Trace file browser file (connected to that database as dba) I find these two inconsistency (doubts):

  1. What is object_id=-1 which has the most number of statements?

  2. When you click on above link thta reproduce no statement, old one remain what confuse me (and probably other users).

If needed may sent trc file to you on request.

Brg

Damir

Now when I return to mine example and changed header width, I was able to to see all data in lower pane (which were few minutes ago unreachable)

So “-1” means all objects … ok

But still there are something “strange” with refreshing lower pane with data and sometimes to focus data in upper panel

8(

this file is really strange. here is picture from that shows Query summary

Statement which run as hell is in most waited sql and even numbers are in collision (lower right pane).

Brg

Damir

My guess is that John is going to need the trace file - it’s early here in Dallas area so he may not be up and working yet. I’d recommend zipping and emailing him the trace file. If it’s too big, then zip and put on ftp server or drop box or soemthing and email him instructions to grab it.

Yes, send me the file please.

The -1 in the “waits by object” tab does NOT mean “all objects”. It means in the trace data, the object id was -1. So, in most (presumably all) cases, Oracle was waiting, but not waiting on an object.

For example:

WAIT #18446741324892192776: nam=‘SQL*Net message to client’ ela= 10 driver id=1413697536 #bytes=1 p3=0 obj#=-1 tim=32135482052072

Sorry, one more reply. If you double-click on the statement under the graph, the Trace File Browser will find the statement in the main tab, where all the statements are listed. Then below it, you can see all of the waits associated with that statement.

My guess is that there are a bunch of non-object-related waits on that statement, like maybe “SQL Net Message from client”, which is the wait that Oracle records when it is waiting for the client side application to send its next command. So maybe the user got up to get some coffee or something.

Guys,

thx for quick reply.

I’ll send file when I get home (here I’m totally firewalled)

:slight_smile:

-1 … hm look at next picture, where i focused (same trace file):

How come SQL*Net had some statements … do not follow this … sorry.

SQL*Net maybe the issue but then this wait column has totally confusing meaning in this scope … confused me at least.

Looking forward to hear more from you after you open trc fiel in your Toad.

Brg

Damir

P.S.

If all is then “logical” from your point of view, I think that Bert could write some blog on this topic-Understand situations in Trace file browser

P.S.

Trace File Browser I use in 95% of traces … just what I need in most of situations.

Sorry, one more reply. If you double-click on the statement under the graph, the Trace File Browser will find the statement in the main tab, where all the statements are listed. Then below it, you can see all of the waits associated with that statement.

I did that. Here are three outputs (do not know if you can get them at your place) which I find interesting for review your statement.

Brg

Damir

How come SQL*Net had some statements …

Because the database was waiting on SQL*Net sometime during or after the statement (but before the next statement)

You can go here in the Oracle docs to see what the meaning of each event is.

docs.oracle.com/…/waitevents.htm

Wait events are associated with a statement. In some cases, the statement won’t complete until the wait is finished. In other cases, the statement is already finished, and the associated statement just happens to be the last statement finished before the wait event.

In the example I have above, #18446741324892192776 refers to the cursor number. So, somewhere in my trace file, I can find a line that says:

Parsing in cursor # 18446741324892192776 (followed by the statement itself).

That’s how I connect waits to statements.

When a COMMIT or ROLLBACK is issued (or some other event that causes a transaction to finish, like DDL), you’ll see lines like this.

XCTEND rlbk=1, rd_only=1, tim=32135479409461

WAIT #0: nam=‘SQL*Net message to client’ ela= 2 driver id=1413697536 #bytes=1 p3=0 obj#=-40016363 tim=32135479409533

WAIT #0: nam=‘SQL*Net message from client’ ela= 575 driver id=1413697536 #bytes=1 p3=0 obj#=-40016363 tim=32135479410150

XCTEND indicates the transaction.

The Wait#0 lines are what appear in the “Transaction Waits” grid. These waits are more related to the ending of the transaction than the prior statement, but I have to put them somewhere, so I put them with the prior statement.

Cary Millsap has a good book called “Optimizing Oracle Performance” that covers understanding of trace files (among other things). I used this book as a reference when writing the Trace File Browser. It has a section that walks you through a trace file line by line, explaining what each line means.

Evening all,

In a trace file wait line, an object id of -1 down indeed mean “no object”.

Cheers,

Norm. [TeamT]

Sent from my Samsung Galaxy Note III - powered by Three

John,

Until now I was never greping raw trc file but tkprof (trca) and lately Toad representation.

All I wanted to say that maybe some guideline about how data are presented maybe worth of writing … just that.

However, not many of us want to dig in raw trc files … so this is why I do love tools that help me a lot

Brg

Damir

If you go here
https://groups.yahoo.com/neo/groups/toadbeta/files you can find a document that I wrote when the TFB was first released. It’s a little out of date but still may be helpful.

From: damir.vadas_531 [mailto:bounce-damirvadas_531@toadworld.com]

Sent: Thursday, March 27, 2014 11:32 AM

To: toadoraclebeta@toadworld.com

Subject: RE: [Toad for Oracle - Beta Discussion Forum] Trace file browser inconsistency (doubts)

RE: Trace file browser inconsistency (doubts)

Reply by damir.vadas_531

John,

Until now I was never greping raw trc file but tkprof (trca) and lately Toad representation.

All I wanted to say that maybe some guideline about how data are presented maybe worth of writing … just that.

However, not many of us want to dig in raw trc files … so this is why I do love tools that help me a lot

Brg

Damir

To reply, please reply-all to this email.

Stop receiving emails on this subject.

Or
Unsubscribe from Toad for Oracle - Beta
notifications altogether.

Toad for Oracle - Beta Discussion Forum

Flag
this post as spam/abuse.