Analyze several traces together

With Analyze together, several trace files become a single analysis, the collection. WaitScope joins the segments of each file in time order and builds a single report, with the total time of the whole operation.

When to use it

• The application changes tracefile_identifier in the middle of the operation. The same session continues in another file, and Oracle marks the switch with *** TRACE CONTINUES IN FILE in one file and *** TRACE CONTINUED FROM FILE in the other. A file can receive segments again later.
• The operation went through several sessions. This is the case of tracing by client_id or by service, with one file per process (turn on by client_id).

Analyzed alone, each file loses the cursors opened in the others and counts as its own the time when the session was writing to another file. Do not combine traces from different operations: the mixed profile describes none of them.

What WaitScope does with the files

1. Cuts each file at the switch marks, TRACE CONTINUES IN FILE and TRACE CONTINUED FROM FILE.
2. Joins the segments from the same instance, the same process and the same session, in tim order.
3. Places different sessions one after the other, by start time, without carrying the open cursors of one session into the next.
4. Leaves out, with a warning, a file identical to another one in the collection and a file that is not a trace.
5. Analyzes the assembled trace as a single trace.

A cursor opened in one file is recognized in the other, and the interval when the session was writing to another file does not count twice. Sessions do not mix: the total time of the collection is the sum of the total time of each session.

How to upload

1. Choose all the files of the operation, the folder that holds them or a zip or tar with all of them.
2. Click Upload and analyze.
3. At the question Are these traces from the same process?, choose Yes, same process: analyze together.

Result: a single analysis in History, with the collection badge next to the name. The name is the common start of the file names with the count, as in zz_lote_etapa (3 arquivos) for zz_lote_etapa1.trc, zz_lote_etapa2.trc and zz_lote_etapa3.trc. With no common start, it is the first name followed by e mais 2.

In the app, the question is the same. On the page without JavaScript, check Analyze together, as a single collection before uploading. In the API, open the upload with "juntar": true (upload through the API). A single file that is not a zip or tar becomes a regular analysis.

Collection limits

• Up to 20,000 files per collection. Above that, the upload is refused with A collection accepts up to 20000 files.
• The assembled trace follows the limit on content after decompression (limits).
• A file with too many segments has the rest joined into its last segment, and the warning at the top of the report says which file it was.
• For a while, the collection takes up twice the space of the files on the server. Without space, the upload is refused with No space on the server.

The Collection card in the report

In the Summary, the collection card says how many files and how many sessions went in, as in Collection: 3 files from 2 sessions, analyzed as a single trace, and has two lists:

• Collection sessions, when there is more than one session: the instance, the process, the SID.serial and the time of each session.
• Collection files: the name of each file, its session, the segments that went in, in time order, and the lines.

The assembly warnings, such as a repeated file or segments that overlap in time, appear at the top of the report. In a collection, the Summary does not show File on server. The sample report comes from a single trace and does not have this card; how to read it is in Collection.

Collection uploaded again

The same set of files, in any order, uploaded again to analyze together opens the collection that already exists. The same files uploaded separately become the analyses of each one (repeated upload).

Common problems

• Time gaps between calls in the Diagnosis. A file may be missing from the collection: check the list in Collection files (gaps between calls).
• Warning about segments that overlap in time. The warning asks you to check that the files belong to the same collection: check that they all came from the same operation.
• no file in the collection looks like an Oracle trace: none of the uploaded files is a 10046 trace (file that is not a trace).

Next steps

• Read the card field by field in Collection.
• To generate the files: mark the name with tracefile_identifier.
• To upload: upload through the browser.